Systematic review of relapse, remission, and recovery definitions in anorexia nervosa research

REVIEW Open Access
What happens after treatment? A
systematic review of relapse, remission, and
recovery in anorexia nervosa
Sahib S. Khalsa1,2*, Larissa C. Portnoff3, Danyale McCurdy-
McKinnon4 and Jamie D. Feusner5
Abstract
Background: Relapse after treatment for anorexia nervosa (AN)
is a significant clinical problem. Given the level of
chronicity, morbidity, and mortality experienced by this
population, it is imperative to understand the driving forces
behind apparently high relapse rates. However, there is a lack of
consensus in the field on an operational definition
of relapse, which hinders precise and reliable estimates of the
severity of this issue. The primary goal of this paper
was to review prior studies of AN addressing definitions of
relapse, as well as relapse rates.
Methods: Data sources included PubMed and PsychINFO
through March 19th, 2016. A systematic review was
performed following the PRISMA guidelines. A total of (N =
27) peer-reviewed English language studies addressing
relapse, remission, and recovery in AN were included.
Results: Definitions of relapse in AN as well as definitions of
remission or recovery, on which relapse is predicated,
varied substantially in the literature. Reported relapse rates
ranged between 9 and 52%, and tended to increase
with increasing duration of follow-up. There was consensus that

risk for relapse in persons with AN is especially
high within the first year following treatment.
Discussion: Standardized definitions of relapse, as well as
remission and recovery, are needed in AN to accelerate
clinical and research progress. This should improve the ability
of future longitudinal studies to identify clinical,
demographic, and biological characteristics in AN that predict
relapse versus resilience, and to comparatively
evaluate relapse prevention strategies. We propose standardized
criteria for relapse, remission, and recovery, for
further consideration.
Keywords: Anorexia nervosa, Treatment, Outcome, Relapse,
Remission, Recovery, Prevention, Eating disorder,
Bulimia nervosa
Plain English Summary
Relapse occurs frequently in individuals receiving treat-
ment for anorexia nervosa. However, there is no com-
mon agreement on how to define relapse. In this study,
we reviewed previous studies of relapse, remission, and
recovery following treatment for anorexia nervosa. We
found that there were many different definitions for
these terms, which resulted in different estimates of re-
lapse rate. To understand what drives relapse it is
important to have a consistent definition across studies.
To help this discussion we propose common criteria for
relapse, remission, and recovery from anorexia nervosa.
Background
Anorexia nervosa (AN) is a serious psychiatric illness
with amongst the highest mortality rates of any mental
disorder—up to 18% in long-term follow-up studies [1–
3]. Most cases emerge during adolescence, and tend to-

wards a protracted and chronic course [4, 5]. In females,
AN has a point prevalence of 0.3–1.0% and lifetime
prevalence of 1.2–2.2% [6]. Treatment often succeeds in
temporarily restoring weight, but AN individuals are at
an exceedingly high risk for early relapse [7], and
* Correspondence: [email protected]
1Laureate Institute for Brain Research, 6655 S Yale Ave, Tulsa,
OK 74136, USA
2Oxley College of Health Sciences, The University of Tulsa,
1215 South
Boulder Ave W, Tulsa, OK 74119, USA
Full list of author information is available at the end of the
article
© The Author(s). 2017 Open Access This article is distributed
under the terms of the Creative Commons Attribution 4.0
International License
(http://creativecommons.org/licenses/by/4.0/), which permits
unrestricted use, distribution, and
reproduction in any medium, provided you give appropriate
credit to the original author(s) and the source, provide a link to
the Creative Commons license, and indicate if changes were
made. The Creative Commons Public Domain Dedication waiver
(http://creativecommons.org/publicdomain/zero/1.0/) applies to
the data made available in this article, unless otherwise stated.
Khalsa et al. Journal of Eating Disorders (2017) 5:20
DOI 10.1186/s40337-017-0145-3
http://crossmark.crossref.org/dialog/?doi=10.1186/s40337-017-
0145-3&domain=pdf
mailto:[email protected]
http://creativecommons.org/licenses/by/4.0/
http://creativecommons.org/publicdomain/zero/1.0/

upwards of 50% relapse within the first year after suc-
cessful hospital treatment [8]. The current lack of robust
and reliable responses to treatment highlights the need
for an improved ability to predict illness trajectories.
The primary focus of this review is on how relapse is
defined following treatment for AN. Since relapse is typ-
ically defined relative to recovery and remission, we also
consider how recovery and remission are defined. Pike
has previously eloquently reviewed relapse, recovery, re-
mission, and response in AN [8]. However, since then 11
studies have addressed this topic. The current review
therefore incorporates these additional publications.
In preparing this review, a lack of clarity and uniform-
ity with regard to how to best define relapse, recovery,
and remission was apparent. This perspective is rein-
forced by a literature review of remission in eating disor-
ders concluding that the definitions and associated rates
vary considerably [9]. Fifteen years ago, a European col-
laboration of experts (COST Action B6) adapted defini-
tions for relapse, recovery, partial and full remission, and
recurrence from the depression literature to AN and bu-
limia nervosa (BN) [10]. Despite rigorous consensus-
building and empirical testing of 233 inpatients with
AN, these criteria have not been uniformly adopted by
the field. To date there are no consensus guidelines
available for clinicians or researchers at the professional
or institutional level providing standardized operational
definitions of relapse, recovery, or remission in AN. This
is limiting. A greater consensus regarding the definition
of these constructs would be of considerable benefit to
clinicians, researchers, patients, and family members, by
allowing all constituents to speak the same language.
We performed a focused review of the extant litera-

ture with the primary aim of examining how these
terms have been defined, in order to improve defini-
tions of relapse, recovery and remission in AN.
Reviewing relapse rates was a secondary goal. We
propose a set of standardized criteria for relapse, re-
covery, and remission from AN, which are internally
cohesive and can facilitate longitudinal assessment by
clinicians and researchers.
Methods
Search and study selection
We conducted a systematic qualitative review according
to the PRISMA guidelines, searching the PubMed and
PsychINFO databases. We used keywords for either “an-
orexia nervosa” or “eating disorders” along with “re-
lapse,” or “recovery,” or “remission.” We used an open
search procedure. We also performed the same searches
on Google Scholar to locate relevant articles that the other
search methods possibly overlooked (none were identi-
fied). Our search covered articles that were published
from 1975 to March 19th, 2016. Titles and abstracts were
evaluated and full text was reviewed for relevant stud-
ies. References sections were screened manually for add-
itional studies unidentified via database search.
Eligibility criteria
Participants had to meet ICD-10, DSM-III, IV, or 5 diag-
nostic criteria for AN for inclusion. Studies (n = 1) focus-
ing on binge eating providing relevant information
regarding relapse risk in AN or treatment outcomes of
AN were also included. Studies examining BN and AN
were included, but not those focused solely on BN (n =
2) (except for one [11] that provided treatment informa-
tion pertinent to AN binge-purge (AN-BP) subtype).

Omitted studies included those focused on unspecified
eating disorders (n = 2), comorbid psychiatric disorders
(n = 2), or those without clinical descriptions of relapse
or recovery (n = 3). Non-English language articles were
excluded (n = 6).
Data review and study quality assessment
Three authors (LCP, SSK, and JF) independently ex-
tracted the following data from the selected studies: first
author, publication year, country, and whether the study
was related to relapse, recovery, or remission. To evalu-
ate the quality of the studies, we performed a systematic
review of each article using the National Heart, Lung,
and Blood Institute Study Quality Assessment Tool [12].
This tool provides a rating checklist for each study type.
Three authors (LCP, DM, SSK) independently evaluated
each study according to the rating checklist, and ren-
dered a rating of “Good” or “Fair” or “Poor.” Study qual-
ity was determined by comparing ratings agreement,
with consensus required among reviewers. Discrepancies
in study quality rating were reconciled via discussion of
the individual items on the ratings checklist to arrive at
consensus agreement on the quality indicator. Disagree-
ments were resolved through discussion and consensus.
There were no biases or poor methods identified that
warranted exclusion from the review.
Results
We identified 27 studies meeting eligibility criteria (see
Fig. 1). An overview of pertinent study characteristics
and definitions of recovery/remission and relapse in AN
are listed in Tables 1 and 2. Definitions of relapse were
fundamental to understanding the reported rates in
these studies. Our review revealed widely varied defini-
tions of relapse and recovery/remission in AN. Defini-
tions of recovery and remission are reviewed first since

relapse is predicated upon them.
Definitions of recovery and remission
Recovery typically requires an extended period of time
during which minimal or no criteria for the disorder are
Khalsa et al. Journal of Eating Disorders (2017) 5:20 Page 2 of
12
met, whereas remission requires a shorter duration [13].
The literature can roughly be divided into articles that
(1) define remission/recovery based solely on weight
measurement, (2) define remission/recovery based solely
on symptom reports, (3) define remission/recovery
based solely on weight and symptom reports, i.e., diag-
nostic criteria available at the time. We briefly review
these studies next (Table 1 lists studies providing defini-
tions of partial remission, full remission, and recovery).
Several studies used body mass index (BMI) as the
only criterion for recovery. Cutoffs included a BMI
above 19 [14] or 20 [7, 15]. In contrast, some described
remission based solely on psychiatric symptoms. In one,
full remission was defined as an absence of all symptoms
or only “residual symptoms” for at least 12 weeks, and
partial remission was defined as a reduction of symp-
toms to a sub-diagnostic level for at least 12 weeks [16].
Adopted from the MacArthur guidelines for depression
[13], Keel et al. [17] defined full remission as a Psychi-
atric Status Rating (PSR) score of ≤2 for 8 weeks. Clau-
sen [18] used the same score for 12 weeks, and defined
partial remission as a PSR ≤3 for 12 weeks.
Other articles described outcomes in terms of body

weight and menstruation, using terminology such as
“good,” “intermediate,” “poor,” or “died” [19–22]. These
criteria, or modifications of them, are often referred to
as the “Morgan-Russell” criteria [19]. A later version
specified remission as weight ≥85% of ideal body weight,
regular menses, and no bingeing or purging behaviors
[23]. Modifying these criteria, recovery was later defined
as not meeting AN DSM-IV-TR criteria for a minimum
of 8 weeks [24].
Several proposed definitions included both weight and
clinical symptoms. Pike [8] defined remission as ≥90% of
ideal body weight, resumption of menses, absence of
compensatory behaviors, and Eating Disorder Examin-
ation (EDE) [25] subscales within 2 standard deviations
(SD) of normal. Recovery was defined as meeting remis-
sion criteria for at least 8 weeks. Strober et al. [4] de-
fined full recovery as the absence of all criteria for at
least 8 weeks, and partial recovery as a “good outcome”
(weight within 15% of average and normal menstruation)
from the Morgan-Russell criteria [19]. Other studies did
not have a duration criterion for the absence of symp-
toms but used the “good outcome” criteria to define re-
covery [20–22]. Stice’s Eating Disorder Diagnostic Scale
defined remission as BMI ≥17.5, regular menses, and no
subthreshold or full threshold eating disorder [26, 27].
Martin [28] defined recovered as having a global rating
scale of “excellent,” meaning an individual was >90%
ideal weight, had regular menstruation, and normal eat-
ing and social patterns. Eckert et al. [29] defined “recov-
ered” as within 15% of ideal body weight, cyclical
menses, and no significant disturbance in eating or
weight control behaviors or body image disturbance.
Kordy et al. [10] defined full recovery for restricting AN
as a BMI >19 and no extreme fear of weight gain for

12 months (plus no purging and no binges for 12 months
Fig. 1 Prisma diagram
12
Table 1 Definitions of recovery and remission, according to
individual studies identified by the literature search
Authors Criteria Duration Study quality
Definitions of Recovery
Martin, 1985 [28] “Excellent”: > 90% of their ideal weight,
regular
menstrual patterns, and eating and social patterns were
normal
Not specified Fair
Norring and Sohlberg, 1993 [34] “Well” defined as having no
eating disorder diagnosis
or remnants of the weight and/or shape preoccupation
Not specified Good
Eckert et al., 1995 [29] ≥85% of ideal body weight, cyclical
menses, and no
significant disturbance in eating or weight control
behavior or body image disturbance
Not specified Good

Strober et al., 1997 [4] Free of all criterion symptoms of
anorexia nervosa or
bulimia nervosa
8 weeks Good
Fichter and Quadflieg, 1999 [21] Outcome “good” defined using
Morgan-Russell criteria Not specified Fair
Pike, 1998 [8] ≥90% of ideal body weight or BMI ≥20,
resumption of
menses, absence of binge eating or compensatory
behaviors, Eating Disorder Examination subscales within
2 SD of normal
8 weeks Fair
Herzog, et al., 1999 [32] Absence of all symptoms or 1–2
residual symptoms—
Psychiatric Status Rating (PRS) score of 1 or 2
8 weeks Good
Lowe et al., 2001 [22] Outcome “good” defined using Morgan-
Russell and PSR 1 Not specified Good
Kordy et al., 2002 [10] AN-R: BMI > 19, no extreme fear of
weight gain
AN-BP: BMI > 19, no extreme fear of weight gain, no
vomiting or laxative abuse, no binges
12 months Good
Carter et al., 2004 [15] BMI above 20 Not specified Good
Walsh et al., 2006 [14] BMI above 19 No information Good

Eisler et al., 2007 [20] Outcome “good” defined using Morgan-
Russell criteria Not specified Good
Bodell and Mayer, 2011 [24] No DSM–IV criteria of AN 8
weeks Fair
Bardone-Cone et al., 2010 [30] Full recovery: BMI ≥ 18.5,
absence of binge-eating,
purging or fasting for at least 3 months, not meeting
criteria for current eating disorder, all EDE-Q subscales
within 1 SD of normal
Partial recovery: same as above, but not needing to
satisfy EDE-Q criterion
Not specified Good
Carter et al., 2012 [7] BMI of 20 and reported no more than one
BP episode
before the end of treatment.
2 weeks BMI and no BP
behaviors over the previous
28 days at the end of treatment
Good
Definitions of Full Remission
Morgan and Hayward, 1988 [23] ≥85% of ideal body weight,
regular menses, and no
binge eating or purging behaviors
Not specified Fair a
Pike, 1998 [8] ≥90% of ideal body weight or BMI ≥20,

resumption of
menses, absence of binge eating or compensatory
behaviors, EDE subscales within 2 SD of normal
Not specified Fair
Stice et al., 2000 [27] BMI ≥17.5, regular menses, and no
current
subthreshold or full threshold eating disorder
Not specified Good a
Kordy et al., 2002 [10] AN-R: BMI > 19, no extreme fear of
weight gain
AN-BP: BMI > 19, no extreme fear of weight gain, no
vomiting or laxative abuse, no binges
12 weeks Good
Keel et al., 2005 [17] Absence of all symptoms or 1–2 residual
symptoms—PSR score ≤2
8 weeks Good
Clausen, 2008 [18] PSR score ≤2 12 weeks Good
Helverskov et al., 2010 [16] Absence of all symptoms/1–2
Residual symptoms—PSR
score of 1 or 2
12 weeks Good
12

Table 1 Definitions of recovery and remission, according to
individual studies identified by the literature search (Continued)
Definitions of Partial Remission
Lowe et al., 2001 [22] Outcome “improved” defined using
Morgan-Russell
criteria and PSR 2, 3, or 4
Not specified Good
Kordy et al., 2002 [10] AN-R: BMI > 17.5
AN-BP: BMI > 17.5 in addition to ≤1 binge per week
and no vomiting or laxative abuse
4 weeks Good
Clausen, 2008 [18] PSR score ≤3 12 weeks Good
Helverskov et al., 2010 [16] PSR score of 3 12 weeks Good
a No NHLBI systematic criteria available to rate this study
type; quality rating reflects consensus agreement between two
rater assessments
Table 2 Definitions of relapse, according to individual studies
identified by the literature search
Authors Criteria Duration Study quality
Definitions of Relapse
Isager et al., 1985 [33] Loss of ≥15% of weight acquired during
course of
treatment (if resulting in weight ≤50 kg)

Any point in time within a 1 year
period
Good
Martin, 1985 [28] If the patient required further psychiatric
treatment
after discharge during follow–up period
Not specified Fair
Norring and Sohlberg, 1993 [34] “Ill” defined as having an
eating disorder Not specified Good
Eckert et al., 1995 [29] Loss of ≥15% of average body weight
(based on
Metropolitan Height-Weight Chart, 1959), after
achieving normal body weight
Any point after achieving normal
weight during inpatient treatment
or the follow up period
Good
Strober et al., 1997 [4] Full (“syndromal”) relapse: weight
<85% of ideal body
weight and recurrence of psychological symptoms
Partial (“subsyndromal”) relapse: recurrence of psycho-
logical symptoms but ≥85% of ideal body weight
Not specified Good
Fichter and Quadflieg, 1999 [21] Outcome “poor” defined using
Morgan-Russell criteria Not specified Fair

Pike 1998 [8] BMI ≤ 18.5 or weight ≤85% of ideal body weight;
a
minimum 1 SD increase on the Eating Disorder Evaluation;
loss of menstrual functioning if it has been previously
normal; increase in restriction leading to weight loss;
and possibly increased binge eating, compensatory
behavior, or associated medical problems
Not specified Fair
Herzog, et al., 1999 [32] Return to full criteria symptoms and/or
Psychiatric
Status Rating (PSR) score of 5 or 6
8 weeks following a state of full
recovery
Good
Lowe et al., 2001 [22] Outcome “poor” defined using Morgan-
Russell criteria
and PSR score of 5 or 6
Not specified Good
Kordy et al., 2002 [10] Change from partial or full remission to
full syndrome
according to DSM-IV
Not specified Good
Carter, et al., 2004 [15] BMI below 17.5 and/or at least one
episode of binge
eating/purging behavior per week
3 consecutive months Good

Keel, et al., 2005 [17] Return to full criteria symptoms and/or
PSR score of 5 or 6 Not specified Good
Walsh et al., 2006 [14] BMI below 16.5 for 2 consecutive
weeks, or severe
medical complications, or risk of suicide, or development
of another psychiatric disorder requiring treatment
2 consecutive weeks (low BMI) Good
Eisler et al., 2007 [20] Outcome “poor” defined using Morgan-
Russell criteria Not specified Good
Clausen, 2008 [18] PSR score ≥3 3 months Good
Bodell and Mayer, 2011 [24] Poor outcome, BMI ≤18.5 (using
modified Morgan-
Russell criteria)
Not specified Fair
Helverskov, et al., 2010 [16] Return to full criteria symptoms
and/or PSR score of 5 or 6 Not specified Good
Carter et al., 2012 [7] BMI < 17.5 or at least one episode of
binge eating/
purging behavior per week
McFarlane et al., 2015 [31] AN: BMI < 18.5
AN-BP: average 4 episodes of bingeing and/or vomiting
per month, or BMI < 18.5

12
for AN-BP). They defined full remission for both sub-
types as meeting the same criteria for 3 months. Partial
remission was a BMI >17.5 and ≤1 binge per week and
no vomiting or laxative abuse for 1 month in AN-BP.
Another proposed definition of full recovery was a BMI
≥18.5, absence of binging, purging, or fasting for at least
3 months, not meeting criteria for a current eating dis-
order, and all EDE-Questionnaire (EDE-Q) subscales
within 1 SD of normal [30]. They defined partial recov-
ery as the same without the EDE-Q criterion.
Definitions of relapse
Different definitions of relapse were identified (see
Table 2). Some definitions were dependent on weight or
BMI measures including: BMI < 16.5 for 2 weeks [14],
and BMI < 17.5 [7, 15] or <18.5 [31] for three consecu-
tive months. Other definitions included 15% loss of aver-
age body weight after achieving normal body weight,
either during the index hospitalization or any time dur-
ing the 10-year follow-up period [29]. Strober et al. [4]
similarly defined relapse as <85% ideal body weight,
which could occur post-discharge or post-recovery. Fur-
thermore, relapse could be partial if the individual had
recurrence of psychological symptoms but sustained
85% of ideal weight, or full relapse if both psychological
symptoms returned and body weight dropped to less
than 85%. Several groups [19–22, 24] defined relapse as
Morgan-Russell criteria of “poor” (BMI ≤18.5).
Other definitions of relapse were dependent on psy-

chiatric symptoms or a combination of psychiatric
symptoms and weight changes. Kordy et al. [10] used
a definition of change from DSM-IV partial or full re-
mission to full syndrome. Clausen [18] defined relapse
as PSR ≥ 3 or PSR ≤ 2 after 3 months remission. Re-
lapse has also been defined as meeting full syndrome
criteria (PSR ≥ 5) after 8 weeks of remission [17, 32]
and after 12 weeks of remission [16]. Pike’s [8] more
in-depth definition of relapse includes weight loss,
EDE increase, medical issues, and a return of disor-
dered eating, whereas Martin’s [28] is the simplest,
requiring only that an individual needs psychiatric
intervention.
Rates of Relapse
Relapse rates of AN were highly variable ranging from a
low of 9% to a high of 52% following treatment, with the
majority of studies reporting rates greater than 25% [4,
7, 10, 14–18, 21, 22, 24, 28, 29, 32–34]. Studies suggest
that adolescents [4, 20, 28] and individuals with restrict-
ing subtype AN [7, 29] have a lower likelihood of re-
lapse. The first year is the most critical, with particular
risk of relapse occurring as early as 3 months post-
treatment [4, 7, 15, 32]. Not surprisingly, those who re-
cover fully have lower relapse rates (9%) than those who
only partially recover (35%) [10]. Together, these results
suggest that while most patients experience brief epi-
sodes of recovery, a large proportion relapse. Moreover,
the risk is particularly high within the first year.
Follow-Up Variability
There was substantial variability in the literature for
follow-up procedures. Initial evaluation time points
ranged from 4 weeks to 17 months post-treatment [4, 7,
14, 15, 17, 20, 28, 32, 35]. Some studies utilized only a

single follow-up time point [15, 28], whereas others
followed patients across multiple time points [4, 7, 14,
17, 20, 32, 35]. Some studies had regular follow-up visits
(e.g., every 4 weeks [14], 3 months [7]), whereas others
had irregularly spaced follow-ups (e.g., 2, 6 and 12 year
follow up [35]).
Variable follow-up intervals could complicate estima-
tions of relapse rates, since relapse rates can vary by dur-
ation of the study follow-up. According to this view,
shorter follow-up durations might be associated with
lower relapse rates than longer durations. We identified
articles supporting this possibility. For example, relapse
in a study measuring at 6 months was lower (9% for fully
recovered and 35% for partially recovered) [10] versus
studies measuring at 1-year (27–70%) [7, 14] (see
Table 3). Relapse rates also varied by remission criteria,
with stricter remission criteria displaying lower relapse
rates than less stringent criteria. This is evidenced by
two 10-year longitudinal studies. Eckert and colleagues
[29] reported higher relapse rates (42%) with less strin-
gent relapse criteria and Strober and colleagues [4] re-
ported lower relapse rates (29.5%) with stricter relapse
criteria.
Discussion
The main finding of this review is that there are almost
as many definitions of relapse, remission, and recovery
as there are studies of them. To help rectify this state of
affairs, we suggest that the eating disorders research and
clinical communities evaluate, test, and ultimately adopt
standardized definitions for relapse, remission, and re-
covery. Depression [13], bipolar disorder [36], and
schizophrenia [37] researchers already utilize standard-
ized definitions of these constructs. Consensus guide-
lines for response, partial response, remission, recovery,

and relapse in obsessive compulsive disorder were also
recently proposed [38]. However, we could identify no
such definitions for AN across organizational websites,
including: the Academy for Eating Disorders, Eating Dis-
orders Research Society, National Eating Disorders As-
sociation, and the European Council on Eating
Disorders.
Standardizing how relapse and recovery are defined in
research could substantially improve our understanding
12
T
a
b
le
3
Ra
te
s
o
f
re
la
p
se
id
en
ti

fie
d
b
y
th
e
lit
er
at
u
re
se
ar
ch
a
A
u
th
o
rs
D
ef
in
it
io
n
Ra
te
Sa
m

p
le
si
ze
Fo
llo
w
u
p
ra
te
St
u
d
y
q
u
al
it
y
Ra
te
s
o
f
Re
la
p
se
Is
ag

er
et
al
.,
19
85
[3
3]
Lo
ss
o
f
≥
15
%
o
f
w
ei
g
h
t
ac
q
u
ire
d
d
u
rin

g
co
u
rs
e
o
f
tr
ea
tm
en
t
(if
re
su
lt
in
g
in
w
ei
g
h
t
≤
50
kg
)
an
y
p

o
in
t
in
ti
m
e
w
it
h
in
a
1
ye
ar
p
er
io
d
26
%
A
N
15
1
N
o
t
re

p
o
rt
ed
at
4–
22
ye
ar
s
(m
ea
n
fo
llo
w
u
p
:1
2.
5
ye
ar
s)
G
o
o
d
N
o

rr
in
g
an
d
So
h
lb
er
g
,1
99
3
[3
4]
“Il
l”
d
ef
in
ed
as
h
av
in
g
an
ea
ti
n
g
d
is

o
rd
er
o
r
d
ea
d
25
%
A
N
48
62
%
at
6
ye
ar
s
G
o
o
d
Ec
ke
rt
et

al
.,
19
95
[2
9]
Lo
ss
o
f
≥
15
%
o
f
av
er
ag
e
b
o
d
y
w
ei
g
h
t
(b
as
ed

o
n
M
et
ro
p
o
lit
an
H
ei
g
h
t-
W
ei
g
h
t
C
h
ar
t,
19
59
),
af
te
r
ac
h
ie
vi

n
g
n
o
rm
al
b
o
d
y
w
ei
g
h
t
42
%
A
N
76
10
0%
at
10
ye
ar
s
G
o

o
d
H
er
zo
g
et
al
.,
19
99
[3
2]
Re
la
p
se
fu
ll
cr
it
er
ia
o
f
sy
m
p
to
m
s

o
r
PS
R
sc
o
re
o
f
5
o
r
6
fo
r
8
w
ee
ks
fo
llo
w
in
g
a
st
at
e
o
f
re
co
ve

ry
40
%
A
N
13
6
93
%
at
7.
5
ye
ar
s
G
o
o
d
Ko
rd
y
et
al
.,
20
02

[1
0]
C
h
an
g
e
fr
o
m
p
ar
ti
al
o
r
fu
ll
re
m
is
si
o
n
to
fu
ll
sy
n
d
ro
m

e
ac
co
rd
in
g
to
D
SM
-IV
9%
A
N
w
h
o
w
er
e
in
fu
ll
re
m
is
si
o
n
/
re
co

ve
ry
;
35
%
A
N
w
h
o
w
er
e
in
p
ar
ti
al
re
m
is
si
o
n
23
3
67
%
at

2.
5
ye
ar
s
G
o
o
d
C
ar
te
r
et
al
.,
20
04
[1
5]
BM
I
b
el
o
w
17
.5
fo
r

3
co
n
se
cu
ti
ve
m
o
n
th
s
o
r
at
le
as
t
o
n
e
ep
is
o
d
e
o
f
b
in
g
e

ea
ti
n
g
/p
u
rg
in
g
b
eh
av
io
r
p
er
w
ee
k
fo
r
3
co
n
se
cu
ti
ve
m
o
n
th
s

fo
llo
w
in
g
a
st
at
e
o
f
re
co
ve
ry
35
%
A
N
51
10
0%
at
0.
5
ye
ar
s
(m
ea

n
fo
llo
w
u
p
:1
.3
±
0.
4
ye
ar
s)
G
o
o
d
C
ar
te
r
et
al
.,
20
12
[7
]
BM

I
b
el
o
w
17
.5
fo
r
3
co
n
se
cu
ti
ve
m
o
n
th
s
o
r
at
le
as
t
o
n
e
ep

is
o
d
e
o
f
b
in
g
e
ea
ti
n
g
/p
u
rg
in
g
b
eh
av
io
r
p
er
w
ee
k
fo
r
3
co
n

se
cu
ti
ve
m
o
n
th
s
fo
llo
w
in
g
a
st
at
e
o
f
re
co
ve
ry
41
%
A
N
o
ve
ra
ll;

70
%
A
N
–
BP
,3
0%
A
N
–
R
10
0
N
o
t
re
p
o
rt
ed
at
1
ye
ar
G
o

o
d
W
al
sh
et
al
.,
20
06
[1
4]
BM
I
b
el
o
w
16
.5
fo
r
2
co
n
se
cu
ti
ve
w

ee
ks
,o
r
se
ve
re
m
ed
ic
al
co
m
p
lic
at
io
n
s,
o
r
ris
k
o
f
su
ic
id
e,
o
r
d

ev
el
o
p
m
en
t
o
f
an
o
th
er
p
sy
ch
ia
tr
ic
d
is
o
rd
er
re
q
u
iri
n
g
tr
ea
tm

en
t
27
%
o
f
th
e
p
la
ce
b
o
g
ro
u
p
an
d
29
%
o
f
th
e
flu
o
xe
ti
n
e
g

ro
u
p
89 (4
4
p
la
ce
b
o
g
ro
u
p
;4
9
flu
o
xe
ti
n
e
g
ro
u
p
)
43
%
at
1
ye

ar
G
o
o
d
Ke
el
et
al
.,
20
05
[1
7]
Fu
ll
cr
it
er
ia
sy
m
p
to
m
s/
PS
R
sc
o
re

o
f
5
o
r
6
36
%
A
N
g
ro
u
p
13
6
93
%
at
9
ye
ar
s
G
o
o
d
H

el
ve
rs
ko
v
et
al
.,
20
10
[1
6]
Fu
ll
cr
it
er
ia
sy
m
p
to
m
s/
PS
R
sc
o
re
o
f
5

o
r
6
19
%
A
N
:f
u
ll
o
r
p
ar
ti
al
re
la
p
se
58
50
%
at
2.
5
ye
ar
s
G
o

o
d
M
ar
ti
n
,1
98
5
[2
8]
“E
xc
el
le
n
t”
:>
90
%
o
f
th
ei
r
id
ea
l
w
ei
g
h

t,
re
g
u
la
r
m
en
st
ru
al
p
at
te
rn
s,
an
d
ea
ti
n
g
an
d
so
ci
al
p
at
te
rn
s
w

er
e
n
o
rm
al
.
“G
o
o
d
”:
so
m
e
b
u
t
n
o
t
in
ca
p
ac
it
at
in
g
d
iff
ic
u
lt

y
in
o
n
e
o
f
th
es
e
ar
ea
s
o
n
ly
9%
o
f
A
N
ad
o
le
sc
en
ts
25
10
0%

at
5
ye
ar
s
Fa
ir
St
ro
b
er
et
al
.,
19
97
[4
]
W
ei
g
h
t
fa
lls
b
el
o
w

85
%
o
f
to
ta
l
b
o
d
y
w
ei
g
h
t/
re
cu
rr
en
ce
o
f
p
sy
ch
o
lo
g
ic
al
sy
m

p
to
m
s
29
.5
%
A
N
p
o
st
-d
is
ch
ar
g
e
re
la
p
se
;
9.
8%
A
N
p
o
st
-p
ar

ti
al
-r
ec
o
ve
ry
re
la
p
se
95
87
%
at
0.
5
ye
ar
s
Fi
n
al
fo
llo
w
u
p
at
15
ye

ar
s
G
o
o
d
Fi
ch
te
r
an
d
Q
u
ad
fli
eg
,1
99
9
[2
1]
O
u
tc
o
m
e
“p
o
o
r”

d
ef
in
ed
u
si
n
g
M
o
rg
an
-R
u
ss
el
l
cr
it
er
ia
20
.8
%
A
N
10
3
98
%

at
6
ye
ar
s
G
o
o
d
Lo
w
e
et
al
.,
20
01
[2
2]
O
u
tc
o
m
e
“p
o
o
r”
d

ef
in
ed
u
si
n
g
M
o
rg
an
-R
u
ss
el
l
cr
it
er
ia
an
d
PS
R
sc
o
re
o
f
5
o
r
6

26
%
A
N
63
90
%
at
21
ye
ar
s
G
o
o
d
Ei
sl
er
et
al
.,
20
07
[2
0]

O
u
tc
o
m
e
“p
o
o
r”
d
ef
in
ed
u
si
n
g
M
o
rg
an
-R
u
ss
el
l
cr
it
er
ia
34
.2

%
A
N
ad
o
le
sc
en
ts
38
95
%
at
5
ye
ar
s
G
o
o
d
C
la
u
se
n
,2
00
8
[1

8]
PS
R
sc
o
re
≥
3
8.
6%
51
69
%
at
2.
5
ye
ar
s
G
o
o
d
Bo
d
el
l
an
d
M

ay
er
,2
01
1
[2
4]
Po
o
r
o
u
tc
o
m
e
BM
I
le
ss
th
an
18
(u
si
n
g
M
o
d
ifi
ed

M
o
rg
an
Ru
ss
el
l
cr
it
er
ia
)
52
%
o
f
A
N
g
ro
u
p
p
o
o
r
o
u
tc
o
m

e
21
86
%
at
o
n
e
ye
ar
Fa
ir
a
N
=
1
4
7
4
p
at
ie
n
ts
in
cl
u
d
ed
ac

ro
ss
al
l
st
u
d
ie
s.
R
ep
o
rt
ed
fo
llo
w
u
p
ra
te
s
va
ri
ed
tr
em
en
d
o

u
sl
y,
fr
o
m
4
3
to
1
0
0
%
ac
ro
ss
0
.5
to
2
1
ye
ar
s.
H
o
w
ev
er
,t
h

e
av
er
ag
e
fo
llo
w
u
p
ra
te
w
as
re
la
ti
ve
ly
h
ig
h
at
8
2
.2
%
12

of the pathophysiology of AN and help ground studies
of efficacy and effectiveness, as argued previously [39,
40]. Consensus would increase the quality of meta-
analytic studies. It would facilitate multi-site compari-
sons, which are necessary to improve statistical power for
studying this relatively rare condition. Precise and consist-
ent terminology would also enhance communication
amongst researchers, clinicians, and caregivers.
We propose a unifying framework with potential defi-
nitions for recovery, remission, and relapse to energize
the discussion (see Fig. 2). These definitions are intern-
ally logical, consistent, and conducive to longitudinal
assessment of AN. We advocate the adoption of stan-
dardized definitions for partial and full recovery and
partial and full relapse. DSM-5 defines partial and full
remission, but not partial or full recovery, and the dur-
ation requirement is vague (“a sustained period”) [41].
We propose that definitions of relapse in AN should en-
compass both clinical symptoms and signs such as BMI
measures,1 as has been proposed for definitions of re-
covery [42], to more comprehensively capture the dis-
order. Importantly, our suggested criteria for recovery,
remission, and relapse include objective measures (BMI;
observable behaviors of restricting, binging, and pur-
ging), subjective measures (fear of gaining weight, dis-
turbance of body image), standardized ratings (EDE),
and specific durations of follow-up (1, 3, 6, and
12 months) that are conducive to utilization across both
clinical and research settings (see Fig. 3).
It is worth noting that the proposed approach shares

certain similarities with previous efforts to identify pat-
terns of recovery in AN. For example, the Psychiatric
Status Rating (PSR) scale represented a single six-item
clinician rating based on DSM-III criteria [43]. Lower
scores on this scale, such as a 1, indicated ‘usual self’ or
the absence of meeting diagnostic criteria, whereas
higher scores, such as a 6, indicated presence of ‘definite
criteria, severe.’ The PSR is similar to our proposed
Fig. 2 Proposed standardized definitions of relapse, remission,
and recovery. These standardized definitions were synthesized
from the different
criteria for relapse, remission, and recovery in individual
studies identified by our systematic review. We include a
graphical representation of
these definitions as a useful heuristic tool for conceptualizing
the major transition points (relapse in red, remission in yellow,
recovery in green)
while at the same time underscoring the continuum of pathology
existing within each stage. Note 1: since weight and height
normally increase
until age 20 in pediatric and adolescent populations, age- and
gender- adjusted BMI percentiles for determining expected body
weight (EBW) are
more appropriate in these subgroups, as demonstrated by [52].
Note 2: determination of ideal body weight is complex, and
subject to consideration
of racial, ethnic, demographic, and cultural factors [53]. Note 3:
Symptoms and behaviors are discrete variables, which are
rated/ascertained by the
clinician based on all available clinical information
12

approach in the sense that both require clinician ratings,
and both load upon features of AN that are relevant to
diagnostic criteria in terms of weight status, symptom
burden, and ongoing behaviors. However, our proposed
criteria diverge principally with respect to (1) a focus on
defining stages of relapse, remission, and recovery, (2)
reliance upon a standardized and clinically validated
interview (EDE), and (3) utilization of terminology (par-
tial or full relapse, partial or full remission, partial or full
recovery) that are transparent and can be utilized uni-
formly with patients, caregivers, and clinicians. Our EDE
cutoff selection for partial relapse (greater than or equal
to 2 SD below normal) is also consistent with the ‘cutoff
point a,’ which as previously suggested by Jacobsen et al.
[44], represents a conservative and stringent approach to
determining clinically significant changes.
Due to the highest risk of relapse being in the first
year [4, 17, 20, 32, 33] and relapse often occurring as
early as 3 months post-treatment [4, 7, 15, 32], we rec-
ommend that longitudinal studies conduct follow up as-
sessments no less than every 3 months for the first year,
and every 6 months thereafter for longer studies. With-
out standardized definitions, a refined understanding of
the specific outcomes posed by putative risk factors, and
guidance on measurement, we are in danger of adding
more variability to this literature. Clinically, standardized
definitions for relapse, remission and recovery, com-
bined with consistent monitoring, would help provide
consistent and relevant feedback to patients and family
members regarding their level of risk.
There are several important limitations to consider
when interpreting this review. There is an inherent diffi-
culty identifying the true risk factors predicting AN re-

lapse given the disparate definitions of relapse and
recovery provided to date, potentially giving our review
the appearance that it is challenged by a lack of synthe-
sis. We argue that this challenge is precisely what future
studies would overcome by adopting and adhering to
one set of standards. Secondly, our interpretations are
restricted to the somewhat obvious conclusions that AN
is: (1) characterized by high relapse rates, (2) that re-
lapse rates increase with follow-up lengths, and (3)
there are few reliable predictors. While it seems nearly
impossible to glean generalizations from such hetero-
geneous findings, this highlights the necessity for con-
sensus and standardized definitions. It is important to
emphasize that while the current review has focused
on AN, based in part, on our own research efforts, we
believe that similar consensus standards are needed for
other eating disorders such as bulimia nervosa, binge
eating disorder, and unspecified eating disorder. Al-
though advancing such definitions are beyond the
scope of our qualitative review, we hope that highlight-
ing this disparity will provoke further discussion and
progress. Finally, adding a meta-analytic approach
could derive ‘quantitative data’ characterizing out-
comes, but at this point, would not be additively in-
formative given the aforementioned limitations. This
approach would be useful for a future analysis of ag-
gregated studies using uniform definitions.
Fig. 3 Illness trajectories across a 2 year time period for three
hypothetical individuals with AN exhibiting different illness
courses. One individual
with an uncomplicated course shows a consistent transition from
full relapse to full remission to full recovery. Another
individual shows a complicated
course marked by partial remission, partial relapse, and partial
recovery, followed by a decline to full remission. A third

individual shows a complicated
course with no recovery marked by intermittent bouts of full
relapse punctuated by partial relapse and partial remission. For
an analogous depiction of
illness trajectory based on actual patients, see Kordy et al., [10]
12
The value of reaching consensus
It will be important to carefully consider the value of
reaching consensus on definitions of relapse, remission,
and recovery, who will benefit, and how a consensus
would be best achieved. It is hard to imagine a lasting
consensus without the support of eating disorder organi-
zations. These include organizations which are science-
oriented (e.g., Eating Disorder Research Society (EDRS)
[45], Academy for Eating Disorders (AED) [46] Euro-
pean Council on Eating Disorders (ECED) [47]),
clinician-oriented (AED, National Eating Disorders As-
sociation (NEDA) [48], and International Association of
Eating Disorders Professionals (IAEDP) [49]), and pa-
tient and caregiver-oriented (e.g., Families Empowered
and Supporting Treatment of Eating Disorders (FEAST)
[50], National Alliance on Mental Illness (NAMI) [51],
AED, and NEDA).
It is also necessary to prospectively consider the po-
tential challenges to achieving a consensus. In this
regard, the highly interdisciplinary perspectives required
in the research and treatment of eating disorders
(pediatrics, family medicine, psychiatry, psychology, nu-
trition and dietetics, social work, licensed therapy and
counseling, and nursing) results in complex and often

diverging multifactorial models, which risks a fracturing
of consensus regarding these conditions.
Concrete suggestions for harmonizing this discussion
include (1) the development of conference symposia, (2)
cross-organization workgroups or task forces, and (3)
the generation of consensus statements focused on the
topic. Other practical considerations include feasibility
assessments. For example, follow up frequency will al-
ways be of concern, and conducting monthly, quarterly,
and perhaps even bi-annual follow-ups requires re-
sources that may be infeasible for certain research
groups. We would argue that follow up assessment oc-
curring at any frequency should use a standardized ap-
proach that is comparable to other laboratories. In-
person assessments might be supplemented by phone in-
terviews, and/or the remote collection of collateral infor-
mation from family members, and we observed evidence
of this pragmatic approach in the literature surveyed in
this paper.
Conclusion
The heterogeneity and severity of AN presentation poses
challenges to understanding why relapse occurs, and
how to prevent it. We posit that the eating disorders
community will benefit from considering, testing, and
adopting standardized definitions for relapse, remission,
and recovery. To galvanize this movement, we have
attempted to provide a unifying framework with internally
logical and consistent definitions. This framework is con-
ducive to longitudinal clinical and research assessment,
not only for AN, but for bulimia nervosa, binge eating dis-
order, unspecified eating disorder, and other eating disor-
ders. Without consensus, uncertainty and variability in the
reported recovery, remission, and relapse rates will persist.

Standardizing definitions in AN is a critical first step in
identifying at-risk individuals, and can ultimately advance
the development and evaluation of treatments for this life-
threatening illness.
Endnotes
1Since weight and height normally increase until age
20 in pediatric and adolescent populations, age- and
gender- adjusted BMI percentiles for determining ex-
pected body weight (EBW) are more appropriate in
these subgroups (see Le Grange et al., [52]).
Abbreviations
AN: Anorexia nervosa; BMI: Body mass index; BN: Bulimia
nervosa;
DSM: Diagnostic and statistical manual of mental disorders;
EDE: Eating
Disorder Examination; EDNOS: Eating disorder not otherwise
specified;
PSR: Psychiatric Status Rating
Acknowledgments
We would like to thank Michael Strober for helpful discussions
and
comments on the manuscript, Courtney Sheen for administrative
support
with performing the literature review, and Francesca Morfini for
assistance
with manuscript retrieval.
Funding
This research was supported by NIMH grant numbers
R01MH093535 and
R01MH105662 to Jamie D. Feusner, and by NIMH grant number
K23MH112949 to Sahib S. Khalsa. Dr. Khalsa also received

support from The
William K. Warren Foundation and a NARSAD Young
Investigator Award.
Availability of data and materials
This review paper was developed on previously published data
that can be
obtained from the original source studies.
Authors’ contributions
JDF, LCP and SSK conceived the research idea, SSK, LCP, DM
and JDF drafted
and edited the manuscript. All authors have read and approved
the final
manuscript before submission.
Competing interests
The authors declare that they have no competing interests.
Consent for publication
Not applicable
Ethics approval and consent to participate
Not applicable
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional
claims in
published maps and institutional affiliations.
Author details
1Laureate Institute for Brain Research, 6655 S Yale Ave, Tulsa,
OK 74136, USA.
2Oxley College of Health Sciences, The University of Tulsa,
1215 South
Boulder Ave W, Tulsa, OK 74119, USA. 3Department of

Clinical Psychology,
Teachers College, Columbia University, 525 W 120th St, New
York, NY 10027,
USA. 4Department of Pediatrics, The University of California
Los Angeles, 757
Westwood Plaza, Los Angeles, CA 90095, USA. 5Department of
Psychiatry and
Biobehavioral Sciences, The University of California Los
Angeles, Semel
Institute of Neuroscience and Human Behavior, 760 Westwood
Plaza, Los
Angeles, CA 90024, USA.
Khalsa et al. Journal of Eating Disorders (2017) 5:20 Page 10
of 12
Received: 28 January 2017 Accepted: 19 April 2017
References
1. Harris EC, Barraclough B. Excess mortality of mental
disorder. Br J Psychiatry.
1998;173:11–53.
2. Sullivan PF. Mortality in anorexia nervosa. Am J Psychiatry.
1995;152(7):1073–4.
3. Arcelus J, Mitchell AJ, Wales J, Nielsen S. Mortality rates in
patients with
anorexia nervosa and other eating disorders. A meta-analysis of
36 studies.
Arch Gen Psychiatry. 2011;68(7):724–31.
4. Strober M, Freeman R, Morrell W. The long-term course of
severe anorexia

nervosa in adolescents: survival analysis of recovery, relapse,
and outcome
predictors over 10-15 years in a prospective study. Int J Eat
Disord. 1997;
22(4):339–60.
5. Steinhausen HC, Grigoroiu-Serbanescu M, Boyadjieva S,
Neumarker KJ,
Winkler Metzke C. Course and predictors of rehospitalization in
adolescent
anorexia nervosa in a multisite study. Int J Eat Disord.
2008;41(1):29–36.
6. Smink FR, van Hoeken D, Hoek HW. Epidemiology of eating
disorders:
incidence, prevalence and mortality rates. Curr Psychiatry Rep.
2012;14(4):
406–14.
7. Carter JC, Mercer-Lynn KB, Norwood SJ, Bewell-Weiss CV,
Crosby RD,
Woodside DB, Olmsted MP. A prospective study of predictors
of relapse in
anorexia nervosa: implications for relapse prevention.
Psychiatry Res. 2012;
200(2-3):518–23.
8. Pike KM. Long-term course of anorexia nervosa: response,
relapse, remission,
and recovery. Clin Psychol Rev. 1998;18(4):447–75.
9. Ackard DM, Richter SA, Egan AM, Cronemeyer CL. What
does remission tell
us about women with eating disorders? Investigating
applications of
various remission definitions and their associations with quality

of life.
J Psychosom Res. 2014;76(1):12–8.
10. Kordy H, Kramer B, Palmer RL, Papezova H, Pellet J,
Richard M, et al.
Remission, recovery, relapse, and recurrence in eating
disorders:
Conceptualization and illustration of a validation strategy. J
Clin Psychol.
2002;58(7):833–46.
11. Fichter MM, Kruger R, Rief W, Holland R, Dohne J.
Fluvoxamine in
prevention of relapse in bulimia nervosa: effects on eating-
specific
psychopathology. J Clin Psychopharmacol. 1996;16(1):9–18.
12. National Heart, Lung, and Blood Institute (NHLBI) Study
Quality Assessment
Tool [cited 2017 March 15]; Available from:
https://www.nhlbi.nih.gov/
health-pro/guidelines/in-develop/cardiovascular-risk-
reduction/tools/.
13. Frank E, Prien RF, Jarrett RB, Keller MB, Kupfer DJ,
Lavori PW, et al.
Conceptualization and rationale for consensus definitions of
terms in major
depressive disorder. Remission, recovery, relapse, and
recurrence. Arch Gen
Psychiatry. 1991;48(9):851–5.
14. Walsh BT, Kaplan AS, Attia E, Olmsted M, Parides M,
Carter JC, et al.
Fluoxetine after weight restoration in anorexia nervosa: a
randomized

controlled trial. JAMA. 2006;295(22):2605–12.
15. Carter JC, Blackmore E, Sutandar-Pinnock K, Woodside
DB. Relapse in
anorexia nervosa: a survival analysis. Psychol Med.
2004;34(4):671–9.
16. Helverskov JL, Clausen L, Mors O, Frydenberg M, Thomsen
PH, Rokkedal K.
Trans-diagnostic outcome of eating disorders: A 30-month
follow-up study
of 629 patients. Eur Eat Disord Rev. 2010;18(6):453–63.
17. Keel PK, Dorer DJ, Franko DL, Jackson SC, Herzog DB.
Postremission
predictors of relapse in women with eating disorders. Am J
Psychiatry. 2005;
162(12):2263–8.
18. Clausen L. Time to remission for eating disorder patients: a
2(1/2)-year follow-
up study of outcome and predictors. Nord J Psychiatry.
2008;62(2):151–9.
19. Morgan HG, Russell GFM. Value of family background and
clinical features
as predictors fo long-term outcome in anorexia-nervosa - 4-year
follow-up
study of 41 patients. Psychol Med. 1975;5(4):355–71.
20. Eisler I, Simic M, Russell GF, Dare C. A randomised
controlled treatment trial
of two forms of family therapy in adolescent anorexia nervosa:
a five-year
follow-up. J Child Psychol Psychiatry. 2007;48(6):552–60.

21. Fichter MM, Quadflieg N. Six-year course and outcome of
anorexia nervosa.
Int J Eat Disord. 1999;26(4):359–85.
22. Lowe B, Zipfel S, Buchholz C, Dupont Y, Reas DL, Herzog
W. Long-term
outcome of anorexia nervosa in a prospective 21-year follow-up
study.
Psychol Med. 2001;31(5):881–90.
23. Morgan HG, Hayward AE. Clinical assessment of anorexia
nervosa. The
Morgan-Russell outcome assessment schedule. Br J Psychiatry.
1988;152(3):
367–71.
24. Bodell LP, Mayer LE. Percent body fat is a risk factor for
relapse in anorexia
nervosa: a replication study. Int J Eat Disord. 2011;44(2):118–
23.
25. Fairburn CG, Cooper Z. The Eating Disorder Examination.
In: Fairburn CG,
Wilson GT, editors. Binge eating: Nature, assessment and
treatment. New
York: Guilford Press; 1993.
26. Stice E, Fisher M, Martinez E. Eating disorder diagnostic
scale: additional
evidence of reliability and validity. Psychol Assess.
2004;16(1):60–71.
27. Stice E, Telch CF, Rizvi SL. Development and validation of
the Eating
Disorder Diagnostic Scale: a brief self-report measure of
anorexia, bulimia,

and binge-eating disorder. Psychol Assess. 2000;12(2):123–31.
28. Martin FE. The treatment and outcome of anorexia nervosa
in adolescents:
a prospective study and five year follow-up. J Psychiatr Res.
1985;19(2-3):
509–14.
29. Eckert ED, Halmi KA, Marchi P, Grove W, Crosby R. Ten-
year follow-up of anorexia
nervosa: clinical course and outcome. Psychol Med.
1995;25(1):143–56.
30. Bardone-Cone AM, Harney MB, Maldonado CR, Lawson
MA, Robinson DP,
Smith R, Tosh A. Defining recovery from an eating disorder:
Conceptualization, validation, and examination of psychosocial
functioning
and psychiatric comorbidity. Behav Res Ther. 2010;48(3):194–
202.
31. McFarlane T, MacDonald DE, Trottier K, Olmsted MP. The
effectiveness of an
individualized form of day hospital treatment. Eat Disord.
2015;23(3):
191–205.
32. Herzog DB, Dorer DJ, Keel PK, Selwyn SE, Ekeblad ER,
Flores AT, et al.
Recovery and relapse in anorexia and bulimia nervosa: a 7.5-
year follow-up
study. J Am Acad Child Adolesc Psychiatry. 1999;38(7):829–
37.
33. Isager T, Brinch M, Kreiner S, Tolstrup K. Death and
relapse in anorexia

nervosa: survival analysis of 151 cases. J Psychiatr Res.
1985;19(2-3):515–21.
34. Norring CE, Sohlberg SS. Outcome, recovery, relapse and
mortality across six
years in patients with clinical eating disorders. Acta Psychiatr
Scand. 1993;
87(6):437–44.
35. Fichter MM, Quadflieg N, Hedlund S. Twelve-year course
and outcome
predictors of anorexia nervosa. Int J Eat Disord. 2006;39(2):87–
100.
36. Tohen M, Frank E, Bowden CL, Colom F, Ghaemi SN,
Yatham LN, et al. The
International Society for Bipolar Disorders (ISBD) Task Force
report on the
nomenclature of course and outcome in bipolar disorders.
Bipolar Disord.
2009;11(5):453–73.
37. Andreasen NC, Carpenter Jr WT, Kane JM, Lasser RA,
Marder SR, Weinberger
DR. Remission in schizophrenia: proposed criteria and rationale
for
consensus. Am J Psychiatry. 2005;162(3):441–9.
38. Mataix-Cols D, de la Cruz LF, Nordsletten AE, Lenhard F,
Isomura K, Simpson
HB. Towards an international expert consensus for defining
treatment
response, remission, recovery and relapse in obsessive-
compulsive disorder.
World Psychiatry. 2016;15(1):80–1.

39. Noordenbos G. When have eating disordered patients
recovered and what
do the DSM-IV criteria tell about recovery? Eat Disord.
2011;19(3):234–45.
40. Von Holle A, Pinheiro AP, Thornton LM, Klump KL,
Berrettini WH, Brandt H,
et al. Temporal patterns of recovery across eating disorder
subtypes. Aust N
Z J Psychiatry. 2008;42(2):108–17.
41. American Psychiatric Association. Diagnostic and statistical
manual of
mental disorders: DSM-5. 5th ed. Washington, DC: American
Psychiatric
Association; 2013.
42. Couturier J, Lock J. What is remission in adolescent
anorexia nervosa? A
review of various conceptualizations and quantitative analysis.
Int J Eat
Disord. 2006;39(3):175–83.
43. Herzog DB, Sacks NR, Keller MB, Lavori PW, von Ranson
KB, Gray HM.
Patterns and predictors of recovery in anorexia nervosa and
bulimia
nervosa. J Am Acad Child Adolesc Psychiatry. 1993;32(4):835–
42.
44. Jacobson NS, Roberts LJ, Berns SB, McGlinchey JB.
Methods for defining
and determining the clinical significance of treatment effects:
description, application, and alternatives. J Consult Clin
Psychol. 1999;
67(3):300–7.

45. Eating Disorders Research Society. [cited 2017 March 15]
Available from:
http://edresearchsociety.org/.
46. Academy for Eating Disorders. [cited 2017 March 15]
Available from: http://
www.aedweb.org.
47. European Council on Eating Disorders. [cited 2017 March
15] Available from:
http://www.eced.co.uk/.
48. National Eating Disorders Association. [cited 2017 March
15] Available from:
http://www.nationaleatingdisorders.org/.
49. The International Association of Eating Disorders
Professionals. [cited 2017
March 15] Available from: http://www.iaedp.com/.
of 12
https://www.nhlbi.nih.gov/health-pro/guidelines/in-
develop/cardiovascular-risk-reduction/tools/
https://www.nhlbi.nih.gov/health-pro/guidelines/in-
develop/cardiovascular-risk-reduction/tools/
http://edresearchsociety.org/
http://www.aedweb.org/
http://www.aedweb.org/
http://www.eced.co.uk/
http://www.nationaleatingdisorders.org/
http://www.iaedp.com/

50. Families Empowered and Supporting Treatment of Eating
Disorders. [cited
2017 March 15] Available from: http://www.feast-ed.org.
51. National Alliance on Mental Illness. [cited 2017 March 15]
Available from:
http://www.nami.org/.
52. Le Grange D, Doyle PM, Swanson SA, Ludwig K, Glunz C,
Kreipe RE.
Calculation of expected body weight in adolescents with eating
disorders.
Pediatrics. 2012;129(2):e438–46.
53. Caprio S, Daniels SR, Drewnowski A, Kaufman FR,
Palinkas LA, Rosenbloom
AL, Schwimmer JB. Influence of race, ethnicity, and culture on
childhood
obesity: implications for prevention and treatment: a consensus
statement
of Shaping America’s Health and the Obesity Society. Diabetes
Care. 2008;
31(11):2211–21.
• We accept pre-submission inquiries
• Our selector tool helps you to find the most relevant journal
• We provide round the clock customer support
• Convenient online submission
• Thorough peer review
• Inclusion in PubMed and all major indexing services
• Maximum visibility for your research
Submit your manuscript at
www.biomedcentral.com/submit
Submit your next manuscript to BioMed Central

and we will help you at every step:
of 12
http://www.feast-ed.org/
http://www.nami.org/
Reproduced with permission of copyright owner.
Further reproduction prohibited without permission.
AbstractBackgroundMethodsResultsDiscussionPlain English
SummaryBackgroundMethodsSearch and study
selectionEligibility criteriaData review and study quality
assessmentResultsDefinitions of recovery and
remissionDefinitions of relapseRates of RelapseFollow-Up
VariabilityDiscussionThe value of reaching
consensusConclusionSince weight and height normally increase
until age 20 in pediatric and adolescent populations, age- and
gender- adjusted BMI percentiles for determining expected body
weight (EBW) are more appropriate in these subgroups (see Le
Grange et al.,
[52]).AbbreviationsAcknowledgmentsFundingAvailability of
data and materialsAuthors’ contributionsCompeting
interestsConsent for publicationEthics approval and consent to
participatePublisher’s NoteAuthor detailsReferences
Cognitive Behaviour therapy, 2016
voL. 45, no. 4, 259–269
http://dx.doi.org/10.1080/16506073.2016.1161663
© 2016 Swedish association for Behaviour therapy
The health preoccupation diagnostic interview: inter-rater

reliability of a structured interview for diagnostic assessment
of DSM-5 somatic symptom disorder and illness anxiety
disorder
Erland Axelssona, Erik Anderssona, Brjánn Ljótssona, Daniel
Wallhed Finnb and Erik
Hedmana,c
aDivision of psychology, Department of Clinical neuroscience,
Karolinska institutet, Stockholm, Sweden;
bCapio psykiatri nacka, Stockholm, Sweden; cosher Center for
integrative Medicine, Department of Clinical
neuroscience, Karolinska institutet, Stockholm, Sweden
Introduction
The fifth edition of the Diagnostic and Statistical Manual of
Mental Disorders (DSM-5;
American Psychiatric Association, 2013) has introduced several
revisions of its predecessor
(DSM-IV; American Psychiatric Association, 2000). One is that
the DSM-IV somatoform
disorders, including hypochondriasis and somatization disorder,
have been abandoned.
There were multiple reasons for this. The somatoform disorders
were considered difficult to
understand, and—despite their narrow criteria—showed a
tendency to overlap (Dimsdale
et al., 2013). Additionally, a common view was that these
disorders overemphasized the
presence of medically unexplained symptoms, i.e. that a medical
explanation for the patient’s
symptoms had to be ruled out for a psychiatric diagnosis to be
made (Dimsdale et al., 2013).
This praxis drew an unnecessarily sharp line between “medical”
and “psychiatric” patients.

ABSTRACT
Somatic symptom disorder (SSD) and illness anxiety disorder
(IAD)
are two new diagnoses introduced in the DSM-5. There is a need
for
reliable instruments to facilitate the assessment of these
disorders.
We therefore developed a structured diagnostic interview, the
Health
Preoccupation Diagnostic Interview (HPDI), which we
hypothesized
would reliably differentiate between SSD, IAD, and no
diagnosis.
Persons with clinically significant health anxiety (n = 52) and
healthy
controls (n = 52) were interviewed using the HPDI. Diagnoses
were
then compared with those made by an independent assessor,
who
listened to audio recordings of the interviews. Ratings generally
indicated moderate to almost perfect inter-rater agreement, as
illustrated by an overall Cohen’s κ of .85. Disagreements
primarily
concerned (a) the severity of somatic symptoms, (b) the
differential
diagnosis of panic disorder, and (c) SSD specifiers. We
conclude that
the HPDI can be used to reliably diagnose DSM-5 SSD and
IAD.
KEYWORDS
health anxiety; illness
anxiety disorder; reliability;
somatic symptom disorder;
somatization

ARTICLE HISTORY
received 24 november 2015
accepted 1 March 2016
CONTACT erland axelsson [email protected]
Supplemental data for this article can be accessed here
http://dx.doi.org/10.1080/16506073.2016.1161663
http://dx.doi.org/10.1080/16506073.2016.1161663
260 E. AxELSSon ET AL.
As a result, some diagnosed with a somatoform disorder (e.g.
hypochondriasis) took offense
at their somatic symptoms “being all in their head” (American
Psychiatric Association,
2013; Dimsdale et al., 2013). It was also hard for clinicians to
determine how many medical
tests were necessary for a somatoform diagnosis (Dimsdale et
al., 2013). These difficulties
were likely among the reasons why the somatoform disorders
were seldom diagnosed in
clinical practice (American Psychiatric Association, 2013;
Dimsdale et al., 2013; Rief &
Martin, 2014).
The DSM-IV somatoform disorders have now been replaced
with the DSM-5 somatic
symptom and related disorders. Of these, somatic symptom
disorder (SSD) and illness
anxiety disorder (IAD) are likely to be most commonly
diagnosed (American Psychiatric
Association, 2013). In SSD the patient has (A) at least one

somatic symptom (e.g. a local-
ized pain) that causes significant distress or functional
impairment. Related to the patient’s
somatic symptoms is an excessive psychological reaction,
involving either (B1) dispro-
portionate interpretations of the somatic symptoms, (B2)
anxiety about health or somatic
symptoms, (B3) excessive time and energy devoted to health or
somatic symptoms, or a
combination of these B criteria. Although it is not necessary
that one and the same somatic
symptom is continuously present, (C) the state of being
symptomatic has to have been
persistent (“typically [for] more than 6 months”; American
Psychiatric Association, 2013).
There are three specifiers for SSD. The (I) with predominant
pain specifier is given when
the patient’s somatic symptoms primarily involve pain. The (II)
persistent specifier is given
in the case of severe symptoms, marked impairment, and a
duration longer than 6 months.
The (III) current severity specifier can be defined as either
mild, moderate, or severe. Mild
implies that only one B criterion (B1, B2, or B3) is fulfilled,
moderate means that at least
two B criteria are fulfilled, and severe means that at least two B
criteria are fulfilled and
that there are either numerous somatic complaints or one very
severe somatic symptom.
The core criterion of illness anxiety disorder (IAD) is (A) “a
preoccupation with having
or acquiring a serious illness” (American Psychiatric
Association, 2013). There are (B) no
or only mild somatic symptoms, and if the patient is likely to

develop a medical condition,
his or her preoccupation with health has to be excessive. The
patient has (C) a high level
of health-related anxiety and is easily worried about his or her
health. Due to this fear, (D)
the patient displays excessive health-related behaviors, such as
repeated symptom-checking
or avoidance of doctor visits. The (E) health preoccupation has
been present for at least
six months and is (F) not better explained by another
psychiatric syndrome. IAD may be
specified as care-seeking type (indicating frequent use of
medical care) or care-avoidant type
(indicating rare use of medical care).
In contrast to the DSM-IV somatoform disorders, the criteria of
SSD and IAD do not
require that the patient’s somatic symptoms are medically
unexplained. Even if the somatic
symptoms that are central to the health preoccupation are
explained by a somatic disease,
the patient may be given a diagnosis of SSD or IAD under the
condition that the patient’s
response to his or her somatic symptoms is significant and
excessive. A patient with recur-
rent dizziness due to poorly managed diabetes might for
example be given a diagnosis of SSD
if the dizziness is coupled with a persistent and disproportional
fear of multiple sclerosis.
Although both SSD and IAD involve a preoccupation with
health, there are also three
important differences between these disorders. Firstly, SSD
presupposes at least one somatic
symptom (medically explained or not), which leads to
significant distress or disruption of

daily life. In contrast, IAD is given when there are no, or only
minimal, somatic symptoms
CognITIvE BEHAvIouR THERAPy 261
related to the health preoccupation. Secondly, whereas IAD
requires health anxiety (defined
here as a fear of having or acquiring a severe illness), SSD—
strictly speaking—does not.
Thirdly, IAD—like DSM-IV hypochondriasis—requires that the
patient’s complaints have
been present for at least six months, whereas the SSD duration
criterion is more loosely
formulated.
According to the DSM-5, roughly 75% of those meeting
diagnostic criteria for DSM-IV
hypochondriasis will now be classified as having SSD, whereas
about 25% will have IAD
(American Psychiatric Association, 2013). SSD is also meant to
replace both DSM-IV som-
atization disorder, undifferentiated somatoform disorder and
pain disorder (Dimsdale et
al., 2013). Under our interpretation, a typical health anxiety
patient either has SSD or IAD,
but not a combination of the two. In the case of an enduring and
excessive fear of severe
illness, the key question is typically whether the patient’s
somatic symptoms are mild enough
to warrant a diagnosis of IAD (rather than SSD). If significant
health anxiety is coupled
with recurring somatic symptoms that are reasonably distinct,
the patient should typically
be given a diagnosis of SSD (but not IAD). In rare cases,

patients with significant health
anxiety might not qualify for either of the two diagnoses. This
could, for example, be the
case if the patient has a persistent fear of severe illness, but
neither somatic symptoms nor
excessive health-related behaviors.
In the DSM-5 field trials, SSD (then referred to as “complex
somatic symptom disorder
revised”) demonstrated substantial inter-rater reliability
(κ = .61), but was never assessed
together with IAD (Freedman et al., 2013). Over and above
these field trials, to our knowl-
edge, only four studies have attempted to study DSM-5 SSD
and/or IAD, but in doing so
have relied exclusively on post hoc classification based on self-
assessment questionnaires
(Bailer et al., 2016; Häuser, Bialas, Welsch, & Wolfe, 2015;
van Dessel, van der Wouden,
Dekker, & van der Horst, 2016; Voigt et al., 2012).
Since structured interviews are known to enhance the reliability
of psychiatric diagnostic
procedures (Basco et al., 2000; Grove, Andreasen, McDonald-
Scott, Keller, & Shapiro, 1981;
Kranzler et al., 1995), there is a need for reliable diagnostic
instruments to aid clinicians
in assessing new diagnoses introduced in the DSM-5. We
therefore aimed to develop and
investigate the psychometric properties of a structured interview
for DSM-5 SSD and IAD.
We hypothesized that this new interview would prove reliable in
differentiating between
SSD, IAD, and no diagnosis (ND).
Methods

Participants
The present study included both participants with severe health
anxiety and healthy controls
as respondents for diagnostic interviews. All respondents were
recruited via newspaper
advertisements and subsequent self-referral via the Internet. The
healthy control respond-
ents (n = 52) were compensated with two lottery tickets for their
participation. Respondents
with severe health anxiety (n = 52) were not compensated in
this manner, but instead applied
for a clinical trial of cognitive behavior therapy. Information
concerning this clinical trial
was sent to health care providers in Stockholm and Gothenburg,
Sweden. The clinical trial
had 214 applicants that were interviewed. From these applicants
we randomly selected a
sample stratified for age so that it would match the age
distribution of the healthy control
group. Participant data is presented in Table 1.
Materials
The Health Anxiety Inventory (HAI) is a well-established self-
report measure of health anx-
iety; primarily of cognitive and emotional factors associated
with DSM-IV hypochondriasis
(Salkovskis, Rimes, Warwick, & Clark, 2002). The HAI
comprises 64 items, each with a scale

of 4 alternatives (e.g. between “I do not worry about my
health.” and “I spend most of my
time worrying about my health.”), rendering a total HAI score
between 0 and 192. The HAI
has good psychometric properties when administered via the
Internet (Hedman et al., 2015).
The Mini-International Neuropsychiatric Interview 6 (MINI;
Sheehan et al., 1998) is
a structured diagnostic interview for the assessment of common
psychiatric disorders,
including anxiety disorders and major depressive disorder. In
the present study, diagnostic
assessments were based on all MINI modules except that for
antisocial personality disorder.
We developed the Health Preoccupation Diagnostic Interview
(HPDI) with the aim of
discriminating between DSM-5 SSD and IAD, and also to
discriminate persons with SSD
or IAD from persons without these disorders (the interview is
presented in Appendix 1).
Items for the HPDI were formulated, discussed, and revised by
clinical psychologists with
extensive experience in assessing and treating DSM-IV
hypochondriasis. A first draft of
the instrument was pilot tested in a small sample of patients
seeking help for psychiatric
disorders in the mental health department of a community health
center.
After this test phase, items were further revised to increase
accuracy and usability.
The interview finally encompassed 42 items, of which 24 were
questions to the respond-
ent and 18 to the interviewer. For each diagnostic criterion, the

general principle was
that the HPDI would first provide the interviewer with one or
more questions to extract
pertinent information from the respondent (e.g. “Have you
recently been worried about
having or developing a serious illness?”, “Which illnesses have
you been afraid of having
or developing?”), and then prompt the interviewer to decide if
the corresponding cri-
terion was met (e.g. “Does the patient show a preoccupation
with having or acquiring
a serious illness?”).
Table 1. Description of respondents.
Note. the p-values are based on Student’s t-test and χ²-test. Cta,
clinical trial applicants; hai, health anxiety inventory; hC,
healthy controls; iaD, illness anxiety disorder; nD, no
diagnosis; SSD, somatic symptom disorder.
aDiagnosis according to the interviewer.
Combined sample
(N = 104)
Clinical trial applicants
(n = 52)
Healthy controls
(n = 52)
CTA vs HC
age, range 18–73 20–69 18–73
age, mean (SD) 40.5 (13.1) 38.5 (12.9) 42.4 (13.1) p = .14
Female, n (%) 76 (73%) 40 (77%) 36 (69%) p = .38

hai, range 10–150 75–150 10–62
hai, mean (SD) 70.2 (41.6) 107.8 (21.6) 32.7 (12.1) p < .01
SSD, n (%)a 42 (40%) 42 (81%) 0 (0%)
p < .01iaD, n (%)a 7 (7%) 7 (14%) 0 (0%)
nD, n (%)a 55 (53%) 3 (6%) 52 (100%)
Procedure
All respondents, both clinical trial applicants and healthy
controls, completed the HAI via
the Internet. They were then asked routine clinical questions
(such as “Do you have—or
have you ever had—a somatic disease out of the ordinary?”) and
assessed for psychiat-
ric syndromes using the MINI followed by the HPDI. This was
done through telephone
interviews, which is a reliable method of diagnostic assessment
(Rohde, Lewinsohn, &
Seeley, 1997). All interviews were recorded using a digital
voice recorder with an external
telephone pick-up microphone. We then applied the procedure
of Skre, Onstad, Torgersen,
and Kringlen (1991), by which an independent assessor blind to
the diagnoses made by the
interviewer listened to the recorded interviews and, using the
HPDI, determined if patients
met diagnostic criteria for SSD or IAD.
In order to ensure that the assessor was blind to previous

diagnoses, the interviewer
was instructed not to inform the respondent of any diagnostic
considerations. As long as
the respondent expressed even a vague tendency to identify with
the diagnostic criteria
probed, the interviewer strived to always complete the full
HPDI, in order not to reveal
any diagnostic decisions. This included assessing most IAD
items regardless of whether an
SSD diagnosis had previously been made or not. To prevent
diagnostic ratings from being
steered by sample-related expectations (e.g. that healthy
controls would not qualify for a
diagnosis), the assessor was also blind to whether respondents
belonged to the clinical trial
applicant sample or the healthy control sample. In order to
ensure this, all audio files were
assigned randomized identification numbers and date stamps
were removed.
Both the interviewer and the assessor had access to the DSM-5
and were instructed to
consult it when needed. If the assessor required information
about the presence of comorbid
psychiatric syndromes (e.g. whether or not a certain respondent
suffered from panic dis-
order) in order to make a decision concerning the presence of
SSD or IAD, relevant results
of the MINI interview were provided (this occurred in 7 of all
104 cases).
There was also a second assessor who listened to the recorded
interviews. This assessor
proposed a combined diagnosis of SSD and IAD in 41 (39%) of
all 104 cases. Combining
SSD and IAD should only be possible in cases where (a) SSD

does not involve significant
health anxiety (a fear of severe illness), but rather some other
form of health preoccupation
(e.g. kinesiophobia due to pain), and (b) there is a significant
health anxiety which is either
unrelated to somatic symptoms or only coupled with mild
somatic symptoms (so that an
IAD diagnosis can be made). Since the present study involved
healthy controls and a sig-
nificantly health anxious sample, SSD unrelated to health
anxiety was expected to be rare,
and the combination of SSD and IAD therefore highly unlikely.
Since the second assessor
clearly misunderstood the DSM-5 diagnostic criteria, data from
this assessor was excluded
from further analysis but is available from the authors on
request.
After completing the reliability study, we made minor
alterations to the HPDI in order
to make it more user-friendly. SSD item 2, SSD item 8, and IAD
item 4 were somewhat
simplified, as their latter parts were made optional and put in
parentheses. We also made
clear that the follow-up question of SSD item 5 should be posed
if the patient’s somatic
symptoms are not know to be explained by a serious disease. As
SSD item 13 was rarely
understood by the respondents, the phrase “During the present
episode, how long have you
felt distressed or hindered” was changed to “How long have you
been concerned.” IAD item
9 was altered in a similar way. In order to enhance the
reliability of the SSD pain specifier,

a new question was added (SSD item 14). Finally, in order to
prevent misunderstandings
similar to that of the second assessor, we added a reminder to
consider IAD in the case of
mild somatic symptoms. We think it highly unlikely that these
minor simplifications will
in any way have a negative impact on the reliability of the
instrument.
Clinicians
All telephone interviews were conducted by the same
interviewer, a resident clinical psy-
chologist with experience of conducting about 120 diagnostic
telephone interviews with
persons seeking treatment for health anxiety. The interviewer
received supervision by a
doctoral-level licensed psychologist specialized in diagnosis
and treatment of health anxiety.
The assessor whose ratings were included in the main analysis
was a psychology student
in the final year of the master level psychology program. During
the project, the assessor
periodically discussed diagnostic principles (e.g. guidelines for
differential diagnosis) with
other members of the research team. However, in order to
ensure assessment integrity,
specific cases were never mentioned.
Statistical analysis
Cohen’s κ was calculated for the inter-rater agreement on SSD
vs. IAD vs. ND cases, using

the SPSS version 22.0 (IBM Corp., Armonk, NY). As omnibus
reliability estimates (such
as the κ) are sometimes hard to interpret (Cicchetti & Feinstein,
1990), we also calculated
(a) the absolute number of cases agreed, (b) the percentage of
cases agreed, (c) the number
of cases agreed per class (e.g. the number of SSD diagnoses that
overlapped), and (d) the
proportion of agreed cases per class (e.g. the percentage of SSD
diagnoses that overlapped).
The latter was done by dividing the number of agreed cases
within each class (e.g. the
number of SSD diagnoses agreed on) by the mean number of
cases assigned to the class
(e.g. the total number of SSD diagnoses made by both raters,
divided by two). Such a ratio
may be thought of as a “positive predictive value” indicating
how likely a diagnosis (e.g. of
SSD) determined by one rater was to be endorsed by the other
rater. Kappa values and the
corresponding indexes of agreement were also calculated for the
inter-rater agreement on
SSD and IAD diagnostic specifiers. Verbal interpretations of
Cohen’s κ followed the cate-
gorical divisions of Landis and Koch (1977). Hence, “poor”
means κ < 0.00, “slight” means
0.00 ≤ κ ≤ 0.20, “fair” means 0.21 ≤ κ ≤ 0.40, “moderate”
means 0.41 ≤ κ ≤ 0.60, “substantial”
means 0.61 ≤ κ ≤ 0.80, and “almost perfect” means
0.81 ≤ κ ≤ 1.00.
Results
Choosing between SSD, IAD, and ND, the interviewer and the
assessor agreed in 95 (91%)
of all 104 cases. No respondent was diagnosed with both SSD

and IAD by the same clini-
cian. Kappa values and proportion of agreed cases are presented
in Table 2. As shown in
Table 3, agreement on diagnosis specifiers was almost perfect
concerning IAD specifiers,
moderate concerning the SSD pain specifier, and slight to fair
concerning the SSD persis-
tence and severity specifiers. Post hoc weighted κ estimates for
the severity specifier were
roughly similar.
With regard to diagnosis, the interviewer and the assessor
disagreed over nine cases. Five
of these were rated as SSD by the interviewer and IAD by the
assessor. This was primarily
due to disagreement about which somatic symptoms should be
considered mild enough
to permit an IAD—rather than an SSD—diagnosis. Three cases
were rated as SSD by the
Table 2. reliability estimates of inter-rater agreement on SSD,
iaD, and nD.
Note. iaD, illness anxiety disorder; nD, no diagnosis; SSD,
somatic symptom disorder.
anot applicable due to division by zero.
bnumber of cases agreed divided by the average total number of
cases per rater. in other words, the likelihood of a diagnosis
by one rater to be endorsed by the other rater (see “Statistical
analysis”).

Combined sample Clinical trial applicants Healthy controls
agreement, total (SSD vs. iaD vs. nD)
Cohen’s κ .85 .59 n/a a
agreed, n (%) 95 (91.3%) 43 (82.7%) 52 (100%)
Disagreed, n (%) 9 (8.7%) 9 (17.3%) 0 (0%)
agreement, positive cases
agreed SSD, n (%b) 34 (88%) 34 (88%) 0 (n/a a)
agreed iaD, n (%b) 6 (67%) 6 (67%) 0 (n/a a)
agreed nD, n (%b) 55 (97%) 3 (67%) 52 (100%)
agreement, SSD vs. non-SSD
agreed, n (%) 95 (91.3%) 43 (82.7%) 52 (100%)
Disagreed, n (%) 9 (8.7%) 9 (17.3%) 0 (0%)
agreement, iaD vs. non-iaD
agreed, n (%) 97 (93.3%) 45 (86.5%) 52 (100%)
Disagreed, n (%) 7 (6.7%) 7 (13.5%) 0 (0%)
agreement, nD vs. non-nD
agreed, n (%) 101 (97.1%) 49 (94.2%) 52 (100%)
Disagreed, n (%) 3 (2.9%) 3 (5.8%) 0 (0%)
Table 3. reliability estimates of inter-rater agreement on SSD
and iaD specifiers.
Note. estimates are based on cases where the interviewer and
the assessor agreed on an SSD or iaD diagnosis. iaD, illness
anxiety disorder; SSD, somatic symptom disorder.

anot applicable due to division by zero.
bnumber of cases agreed divided by the average total number of
cases per rater. in other words, the likelihood of a specifier
by one rater to be endorsed by the other rater (see “Statistical
analysis”).
Specifier Cohen’s κ Agreed per choice, n (%) Agreed per class,
n (%b)
SSD (n = 34)
Mild .08 11 (32%) 1 (18%)
vs Moderate 4 (29%)
vs Severe 6 (39%)
predominantly pain .45 30 (88%) 2 (50%)
vs not predominantly pain 28 (93%)
persistent .30 24 (71%) 5 (50%)
vs not persistent 19 (79%)
iaD (n = 6)
Care-seeking 1 6 (100%) 4 (100%)
vs Care avoidant 2 (100%)
vs no specifier 0 (n/aa)
interviewer and ND by the assessor. Two of these were cases of
panic disorder, where
disagreement arose as to whether an additional diagnosis of
SSD should be made or not,
and one case mainly had to do with the clinicians disagreeing on

whether the patient’s
health anxiety returned with sufficient frequency to warrant a
diagnosis. The remaining
disagreement—where the interviewer suggested IAD and the
assessor SSD—stemmed from
different views on whether the somatic symptoms of the
respondent were frequent enough
to warrant an SSD—rather than an IAD—diagnosis.
In the clinical trial applicant sample, 11 out of 52 respondents
(21%) reported presently
having at least one organic disease “out of the ordinary.” The
only conditions that were
referred to by more than one respondent were hypothyroidism
(n = 3) and fibromyalgia
syndrome (n = 2). Other medical conditions included a benign
brain tumor, hiatus hernia,
irritable bowel syndrome, immunodeficiency, an unspecified
benign brain malformation,
spinal disc herniation, severe asthma, severe migraine, and
vulvar vestibulitis. Among the
healthy controls, only one respondent reported having a medical
condition worthy of men-
tioning (hypothyroidism).
Discussion
This study investigated the psychometric properties of the
Health Preoccupation Diagnostic
Interview (HPDI): a new structured interview for diagnosing
DSM-5 SSD and IAD. The
results suggest that the HPDI can be a reliable diagnostic
instrument for assessing health
anxiety in terms of SSD and IAD. An important strength of this
study was that the clinicians
who assigned diagnoses did not undergo any special training to

increase rating concordance.
This means that the presented estimates are likely to mirror
what would be attainable in
clinical practice. To our knowledge, this study is the first to
show that these new DSM-5
diagnoses can be simultaneously assessed with acceptable inter-
rater reliability.
The overall (SSD vs. IAD vs. ND) agreement was almost perfect
for the combined sample,
moderate for the clinical sample, and perfect (κ not applicable)
for the healthy controls.
Notably, as the assessor was blind to respondent recruitment
procedures, sample differ-
ences in diagnostic ratings cannot be explained by mere
expectancy: e.g. that the assessor
expected non-clinical respondents not to meet diagnostic
criteria. The results are in line
with previous reliability estimates for many diagnoses made
with the Structured Clinical
Interview for DSM-IV Axis I Disorders (SCID I; Lobbestael,
Leurgans, & Arntz, 2011; Skre
et al., 1991) but slightly lower than the reliability estimates of
most MINI modules (Sheehan
et al., 1998). Additional standardization of the assessment
procedure could possibly further
enhance the psychometric properties of the HPDI. In the present
study, disagreement as to
whether a diagnosis of SSD or IAD should be made primarily
arose due to different views on
how severe certain somatic symptoms were. Therefore, if more
precise definitions of “mild”
or “minimal” somatic symptoms were established, many
disagreements could probably be
avoided. For research purposes, the use of multiple co-operating
raters and/or introductory

rating concordance checks could also be used to further enhance
reliability figures.
Although there was almost perfect agreement on the IAD
specifiers, i.e. care seeking or
care avoidant subtypes, and moderate agreement on the SSD
pain specifier, there was much
higher disagreement on the SSD severity and persistence
specifiers. We see two likely reasons
for this. Firstly, as pointed out by Rief and Martin (2014), the
names of these specifiers are
surprising. Most strikingly, as SSD criterion C ensures that
most SSD patients have had
their symptoms for six months or longer, the persistence
specifier is primarily a measure of
symptom severity and functional impairment. That is, the
persistence specifier is—at least in
most cases—not a measure of persistency at all. Similarly, the
SSD severity specifier seemed
only vaguely related to the respondent’s degree of suffering and
impairment.
Secondly, expanding on our previous suggestions for further
operationalization of “mild”
somatic symptoms, what exactly is a “severe” or “very severe”
somatic symptom? And how
many are “multiple” somatic complaints? We deliberately
refrained from making such clar-
ifications, in order not to arbitrarily divert from the DSM-5
criteria. Unfortunately, these
criteria do not seem to in and of themselves provide sufficient

guidance for the assessment
of the above-mentioned specifiers. Simply put, we suspect that
the SSD severity and persis-
tence specifiers, as phrased in DSM-5, are too vague and
counterintuitive to be sufficiently
reliable. In future revisions of the DSM, we strongly suggest
that these specifiers be better
operationalized.
The primary limitation of this study was that only two
clinicians’ assessments could be
compared. Even though the assessor was blind to the respondent
recruitment procedure,
the interviewer was not, and might therefore—although
instructed not to do so—have
steered the interview based on expectancy (so that individuals
of the non-clinical sample
were treated differently than individuals of the clinical trial
applicant sample). There is also
a possibility that reliability figures had been different (either
better or worse) if the assessor
had conducted separate diagnostic interviews instead of
listening to audio files.
Even though the structured diagnostic interview covered many
common differential
diagnoses of SSD and IAD (such as major depressive disorder,
panic disorder, generalized
anxiety disorder, and obsessive-compulsive disorder), several
conditions of the DSM were
not surveyed in such a systematic manner. Of particular interest,
the remaining somatic
symptom and related disorders (e.g. conversion disorder and the
psychological factors
affecting other medical conditions category), as well as sexual
dysfunction disorders (e.g.

genito-pelvic pain/penetration disorder) were not routinely
assessed using separate prompts
or modules.
There were also limitations with regard to sampling. Since the
two samples of this study
(health anxious versus non-clinical respondents) were recruited
through different channels,
it is likely that the clinicians’ (the interviewer’s and the
assessor’s) ability to distinguish
between diagnostic (SSD or IAD) cases and ND cases was
slightly overestimated. Since
IAD was more rare than SSD and ND, reliability estimates for
IAD are also likely less gen-
eralizable than those for SSD and ND. The presented findings
nevertheless indicate that
the HPDI shows great promise in diagnosing both SSD and IAD.
Although the HPDI likely makes the diagnostic procedure
easier, it is worth underscoring
that structured diagnostic interviews (like the HPDI) demand
that their user is familiar with
the diagnoses being assessed (in this case DSM-5 SSD and
IAD). This may seem obvious,
but was made even more evident by the misunderstanding on
behalf of one of two assessors
in the present study (see “Procedure”). The criteria of SSD and
IAD may seem counterintu-
itive, not least since most clinicians are not used to classifying
somatic symptoms in terms
of severity. Nevertheless, as we have seen, determining somatic
symptom severity is often
important to differentiate SSD from IAD. A thorough
understanding of common differential
diagnoses, such as anxiety and obsessive-compulsive spectrum
disorders, is also required

in order to make a valid diagnosis.
We conclude that the HPDI can be a reliable instrument for
diagnosing DSM-5 SSD and
IAD. It should, however, only be used by clinicians familiar
with the DSM-5, and is not suit-
able for assessing the SSD severity and persistence specifiers.
The psychometric properties of
the HPDI should preferably be further investigated in additional
patient groups (e.g. mixed
psychiatric samples, patients with chronic pain, patients
suffering from severe illness, and
samples involving more patients with IAD). Since the
completion of this study, modules for
SSD and IAD have been included in the Structured Clinical
Interview for DSM-5 (SCID-
5; American Psychiatric Association, 2015) and Anxiety and
Related Disorders Interview
Schedule for DSM-5 (ADIS-5; Brown & Barlow, 2014), but the
psychometric properties
of these modules have—to our knowledge—not yet been
evaluated. A promising SSD B
criteria symptom questionnaire has also been developed by
Toussaint et al. (2016). We
warmly welcome all efforts that help shed further light on SSD,
IAD, and the relationship
between these disorders. Free use of the HPDI is encouraged in
order to achieve this end.
Authors’ contributions
Conception and design: primarily EAx and EH. Procedure and

data acquisition: EAx,
EAn, BL, and EH. All authors—EAx, EAn, BL, DWF, and EH—
contributed significantly
to the analysis, interpretation, and writing process. All authors
read and approved the final
manuscript.
Acknowledgment
We thank MSc Rebecka Bratt for her skillful and invaluable
contributions to this project.
Disclosure statement
The authors declare that they have no conflicts of interest.
Funding
This research was funded by Karolinska Institutet and
Stockholm County Council, which had no
role in the design, realization, or publication of the study.
References
American Psychiatric Association. (2000). Diagnostic and
statistical manual of mental disorders:
DSM-IV-TR. Washington, DC: Author.
American Psychiatric Association. (2013). Diagnostic and
statistical manual of mental disorders:
DSM-5. Arlington, VA: Author.
American Psychiatric Association. (2015). Comparison of
diagnostic coverage of the SCID-5-
RV vs. SCID-5-CV. Retrieved from
https://www.appi.org/File%20Library/Products/AH1522-

Comparison-SCID-RV-vs-CV.pdf
Bailer, J., Kerstner, T., Witthöft, M., Diener, C., Mier, D., &
Rist, F. (2016). Health anxiety and
hypochondriasis in the light of DSM-5. Anxiety, Stress, &
Coping, 29, 219–239. doi:10.1080/106
15806.2015.1036243
Basco, M. R., Bostic, J. Q., Davies, D., Rush, A. J., Witte, B.,
Hendrickse, W., & Barnett, V. (2000).
Methods to improve diagnostic accuracy in a community mental
health setting. American Journal
of Psychiatry, 157, 1599–1605.
http://dx.doi.org/10.1080/10615806.2015.1036243
http://dx.doi.org/10.1080/10615806.2015.1036243
Brown, T. A., & Barlow, D. H. (2014). Anxiety and related
disorders interview schedule for DSM-5
(ADIS-5) – Adult version. New York, NY: Oxford University
Press.
Cicchetti, D. V., & Feinstein, A. R. (1990). High agreement but
low kappa: II. Resolving the paradoxes.
Journal of Clinical Epidemiology, 43, 551–558.
doi:10.1016/0895-4356(90)90159-M
Dimsdale, J. E., Creed, F., Escobar, J., Sharpe, M., Wulsin, L.,
Barsky, A., … Levenson, J. (2013).

Somatic symptom disorder: An important change in DSM.
Journal of Psychosomatic Research,
75, 223–228. doi:10.1016/j.jpsychores.2013.06.033
Freedman, R., Lewis, D. A., Michels, R., Pine, D. S., Schultz,
S. K., Tamminga, C. A., … Oquendo,
M. A. (2013). The initial field trials of DSM-5: New blooms and
old thorns. American Journal of
Psychiatry, 170(1), 1–5.
Grove, W. M., Andreasen, N. C., McDonald-Scott, P., Keller,
M. B., & Shapiro, R. W. (1981). Reliability
studies of psychiatric diagnosis: Theory and practice. Archives
of General Psychiatry, 38, 408–413.
Hedman, E., Ljótsson, B., Andersson, E., Andersson, G.,
Lindefors, N., Rück, C., … Lekander,
M. (2015). Psychometric properties of Internet-administered
measures of health anxiety: An
investigation of the health anxiety inventory, the illness attitude
scales, and the Whiteley index.
Journal of Anxiety Disorders, 31, 32–37.
Häuser, W., Bialas, P., Welsch, K., & Wolfe, F. (2015).
Construct validity and clinical utility of current
research criteria of DSM-5 somatic symptom disorder diagnosis
in patients with fibromyalgia
syndrome. Journal of Psychosomatic Research, 78, 546–552.
Kranzler, H. R., Kadden, R. M., Burleson, J. A., Babor, T. F.,
Apter, A., & Rounsaville, B. J. (1995).
Validity of psychiatric diagnoses in patients with substance use
disorders: Is the interview more
important than the interviewer? Comprehensive Psychiatry, 36,
278–288.

Landis, J. R., & Koch, G. G. (1977). The measurement of
observer agreement for categorical data.
Biometrics, 33, 159–174. doi:10.2307/2529310
Lobbestael, J., Leurgans, M., & Arntz, A. (2011). Inter-rater
reliability of the structured clinical
interview for DSM-IV axis I disorders (SCID I) and axis II
disorders (SCID II). Clinical Psychology
& Psychotherapy, 18, 75–79.
Rief, W., & Martin, A. (2014). How to use the new DSM-5
somatic symptom disorder diagnosis in
research and practice: A critical evaluation and a proposal for
modifications. Annual Review of
Clinical Psychology, 10, 339–367. doi:10.1146/annurev-clinpsy-
032813-153745
Rohde, P., Lewinsohn, P. M., & Seeley, J. R. (1997).
Comparability of telephone and face-to-face
interviews in assessing axis I and II disorders. American
Journal of Psychiatry, 154, 1593–1598.
doi:10.1176/ajp.154.11.1593
Salkovskis, P. M., Rimes, K., Warwick, H., & Clark, D. (2002).
The health anxiety inventory:
Development and validation of scales for the measurement of
health anxiety and hypochondriasis.
Psychological Medicine, 32, 843–853.
doi:10.1017/S0033291702005822
Sheehan, D. V., Lecrubier, Y., Sheehan, K. H., Amorim, P.,
Janavs, J., Weiller, E., … Dunbar, G.
C. (1998). The mini-international neuropsychiatric interview
(M.I.N.I.): The development and
validation of a structured diagnostic psychiatric interview for
DSM-IV and ICD-10. Journal of

Clinical Psychiatry, 59, 22–33; quiz 34–57.
Skre, I., Onstad, S., Torgersen, S., & Kringlen, E. (1991). High
interrater reliability for the structured
clinical interview for DSM-III-R Axis I (SCID-I). Acta
Psychiatrica Scandinavica, 84, 167–173.
Toussaint, A., Murray, A. M., Voigt, K., Herzog, A., Gierk, B.,
Kroenke, K., … Lowe, B. (2016).
Development and validation of the somatic symptom disorder-B
criteria scale (SSD-12).
Psychosomatic Medicine, 78, 5–12.
doi:10.1097/psy.0000000000000240
van Dessel, N. C., van der Wouden, J. C., Dekker, J., & van der
Horst, H. E. (2016). Clinical value of
DSM IV and DSM 5 criteria for diagnosing the most prevalent
somatoform disorders in patients
with medically unexplained physical symptoms (MUPS).
Journal of Psychosomatic Research, 82,
4–10. doi:10.1016/j.jpsychores.2016.01.004.
Voigt, K., Wollburg, E., Weinmann, N., Herzog, A., Meyer, B.,
Langs, G., & Löwe, B. (2012). Predictive
validity and clinical utility of DSM-5 somatic symptom
disorder—Comparison with DSM-IV
somatoform disorders and additional criteria for consideration.
Journal of Psychosomatic Research,
73, 345–350.
http://dx.doi.org/10.1016/0895-4356(90)90159-M
http://dx.doi.org/10.1016/j.jpsychores.2013.06.033
http://dx.doi.org/10.2307/2529310
http://dx.doi.org/10.1146/annurev-clinpsy-032813-153745
http://dx.doi.org/10.1176/ajp.154.11.1593
http://dx.doi.org/10.1017/S0033291702005822

http://dx.doi.org/10.1097/psy.0000000000000240
http://dx.doi.org/10.1016/j.jpsychores.2016.01.004
Copyright of Cognitive Behaviour Therapy is the property of
Routledge and its content may
not be copied or emailed to multiple sites or posted to a listserv
without the copyright holder's
express written permission. However, users may print,
download, or email articles for
individual use.
AbstractIntroductionMethodsParticipantsMaterialsProcedureCli
niciansStatistical analysisResultsDiscussionAuthors’
contributionsAcknowledgmentDisclosure
statementFundingReferences
Cognitive Behaviour therapy, 2016
voL. 45, no. 4, 259–269
http://dx.doi.org/10.1080/16506073.2016.1161663
© 2016 Swedish association for Behaviour therapy
The health preoccupation diagnostic interview: inter-rater
reliability of a structured interview for diagnostic assessment
of DSM-5 somatic symptom disorder and illness anxiety
disorder
Erland Axelssona, Erik Anderssona, Brjánn Ljótssona, Daniel
Wallhed Finnb and Erik
Hedmana,c
aDivision of psychology, Department of Clinical neuroscience,
Karolinska institutet, Stockholm, Sweden;
bCapio psykiatri nacka, Stockholm, Sweden; cosher Center for

integrative Medicine, Department of Clinical
neuroscience, Karolinska institutet, Stockholm, Sweden
Introduction
The fifth edition of the Diagnostic and Statistical Manual of
Mental Disorders (DSM-5;
American Psychiatric Association, 2013) has introduced several
revisions of its predecessor
(DSM-IV; American Psychiatric Association, 2000). One is that
the DSM-IV somatoform
disorders, including hypochondriasis and somatization disorder,
have been abandoned.
There were multiple reasons for this. The somatoform disorders
were considered difficult to
understand, and—despite their narrow criteria—showed a
tendency to overlap (Dimsdale
et al., 2013). Additionally, a common view was that these
disorders overemphasized the
presence of medically unexplained symptoms, i.e. that a medical
explanation for the patient’s
symptoms had to be ruled out for a psychiatric diagnosis to be
made (Dimsdale et al., 2013).
This praxis drew an unnecessarily sharp line between “medical”
and “psychiatric” patients.
ABSTRACT
Somatic symptom disorder (SSD) and illness anxiety disorder
(IAD)
are two new diagnoses introduced in the DSM-5. There is a need
for
reliable instruments to facilitate the assessment of these
disorders.
We therefore developed a structured diagnostic interview, the
Health
Preoccupation Diagnostic Interview (HPDI), which we

hypothesized
would reliably differentiate between SSD, IAD, and no
diagnosis.
Persons with clinically significant health anxiety (n = 52) and
healthy
controls (n = 52) were interviewed using the HPDI. Diagnoses
were
then compared with those made by an independent assessor,
who
listened to audio recordings of the interviews. Ratings generally
indicated moderate to almost perfect inter-rater agreement, as
illustrated by an overall Cohen’s κ of .85. Disagreements
primarily
concerned (a) the severity of somatic symptoms, (b) the
differential
diagnosis of panic disorder, and (c) SSD specifiers. We
conclude that
the HPDI can be used to reliably diagnose DSM-5 SSD and
IAD.
KEYWORDS
health anxiety; illness
anxiety disorder; reliability;
somatic symptom disorder;
somatization
ARTICLE HISTORY
received 24 november 2015
accepted 1 March 2016
CONTACT erland axelsson [email protected]
Supplemental data for this article can be accessed here
http://dx.doi.org/10.1080/16506073.2016.1161663
http://dx.doi.org/10.1080/16506073.2016.1161663

Systematic review of relapse, remission, and recovery definitions in anorexia nervosa research

Systematic review of relapse, remission, and recovery definitions in anorexia nervosa research

Recommended

Recommended

More Related Content

Similar to Systematic review of relapse, remission, and recovery definitions in anorexia nervosa research

Similar to Systematic review of relapse, remission, and recovery definitions in anorexia nervosa research (20)

More from michael591

More from michael591 (20)

Recently uploaded

Recently uploaded (20)

Systematic review of relapse, remission, and recovery definitions in anorexia nervosa research