Lindstedt et al.

Journal of Eating Disorders (2017) 5:4

DOI 10.1186/s40337-017-0135-5

RESEARCH ARTICLE Open Access

Adolescents with full or subthreshold

anorexia nervosa in a naturalistic sample –
characteristics and treatment outcome
Katarina Lindstedt* , Lars Kjellin and Sanna Aila Gustafsson

Abstract
Background: Anorexia Nervosa (AN) destroys developmentally important early years of many young people and
knowledge is insufficient regarding course, treatment outcome and prognosis. Only a few naturalistic studies have
been conducted within the field of eating disorder (ED) research. In this naturalistic study we included adolescents
with AN or subthreshold AN treated in outpatient care, and the overall aim was to examine sample characteristics
and treatment outcome. Additional aims were to examine potential factors associated with remission as an outcome
variable, and possible differences between three time periods for treatment onset.
Methods: Participants were identified through the Swedish national quality register for eating disorder treatment
(SwEat), in which patients are registered at treatment onset and followed up once a year until end of treatment (EOT).
Inclusion criteria were: medical or self-referral to one of the participating treatment units between 1999 and
2014, 13–19 years of age at initial entry into SwEat and diagnosed with AN or subthreshold AN. The total
sample consisted of 3997 patient from 83 different treatment units.
Results: The results show that 55% of the participants were in remission and approximately 85% were within
a healthy weight range at EOT. Of those who ended treatment according to plan, 70% were in remission and
90% within a healthy weight range. The average treatment duration was approximately 15 months. About
one third of the patients terminated treatment prematurely, which was associated with a decreased chance
of achieving remission. Remission rates and weight recovery increased over time, while treatment duration
decreased. Considering treatment outcome, the results did not show any differences between patients with
AN or subthreshold AN.
Conclusions: The present study shows a relatively good prognosis for adolescent patients with AN or subthreshold
AN in routine care and the results indicate that treatment for adolescents with ED in Sweden has become
more effective over the past 15 years. The results of the present study contribute to the scope of treatment
research and the large-scale naturalistic setting secures the generalizability to a clinical environment. However,
more research is needed into different forms of evidence, new research strategies and diversity of treatment approaches.
Trial registration: Registered in FOU in Sweden (Researchweb.org) 2014-04-14, ID nr 147301.
Keywords: Adolescents, Anorexia nervosa, Eating disorders, Naturalistic sample, Treatment

* Correspondence: [email protected]
University Health Care Research Center, Faculty of Medicine and Health,
Örebro University, SE-701 82 Örebro, Sweden

© 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.
Lindstedt et al. Journal of Eating Disorders (2017) 5:4 Page 2 of 10

Plain english summary Naturalistic studies can add valuable knowledge about the
This study is one of few studies within the field of eating impact of experimental results and outcome of various
disorder (ED) research that is conducted in a natural treatments in real life settings, and about descriptive base-
treatment setting. In this study we included adolescents line data for patients. In this naturalistic study we included
with Anorexia Nervosa (AN) or comparable symptoms adolescents with AN or subthreshold AN, according to
treated in outpatient care, and the overall aim was to the Diagnostic and Statistical Manual of Mental Disorders
examine the characteristics of the sample and treatment IV (DSM-IV) [21]. We based this on results from previous
results. Additional aims were to examine possible treat- studies, showing that patients with subthreshold AN most
ment factors that could be associated with being free often suffer from symptoms to the same extent as patients
from an ED diagnosis at end of treatment (EOT), and with AN, despite a higher BMI (Body Mass Index, kg/m2)
possible differences between three time periods for start in general [22–24]. This is also in line with the updated
of treatment. Participants were identified through the criteria in Diagnostic and Statistical Manual of Mental
Swedish national quality register for eating disorder Disorders 5 (DSM-5) [25], in which the definition of AN
treatment (SwEat) and a total of 3997 patients from 83 has been broadened.
different treatment units were included. The results The overall aim of this study was to examine sample
show that 55% of the participants were free from an ED characteristics and treatment outcome in a naturalistic
diagnosis at EOT and about 85% were within a healthy sample of adolescents with AN or subthreshold AN,
weight range. The results show no differences between treated in outpatient care. Additional aims were to
patients with AN or with comparable symptoms, but ter- examine potential factors associated with remission as
minating treatment prematurely imply a decreased chance an outcome variable, and possible differences in sample
of achieving remission. The results indicate that treatment characteristics and treatment outcome between three
has become more effective over the past 15 years. time periods for treatment onset.

Background Methods
Anorexia Nervosa (AN) is a severe form of eating disorder Participants in the present study were identified through
(ED) that is costly, both in terms of personal suffering and the Swedish national quality register for eating disorder
health economy on an individual and societal level, and it treatment (SwEat). SwEat is a longitudinal internet-
destroys developmentally important early years for many based quality assurance register, developed in 1999, that
adolescent girls and boys [1, 2]. Although people of all includes all specialist ED units in Sweden and a fair
ages are affected, AN often has its onset during adoles- number of general psychiatric units. A total of 108 units
cence and mainly affects girls between 15 and 19 years participated in SwEat between 1999 and 2014. The ob-
[3–5]. Despite recent advances within the ED research jectives of SwEat is to document clinically important key
field, there is still inadequate knowledge about the course, variables, such as waiting time, treatment duration, dif-
treatment outcome and prognosis of adolescent AN [6]. ferent types of treatment interventions (e.g. outpatient,
Early treatment interventions have been shown to be day patient or inpatient treatment) and treatment out-
important for the best effects [1, 7, 8]. When treatment is come [26]. Information is registered in SwEat when it is
delayed, the risk increases for more severe and prolonged established that the patient has an ED diagnosis, the unit
symptoms [2, 9]. However, the treatment often progresses intends to treat the patient and the patient has given
slowly [10] and is characterized by high dropout rates her/his consent to registration. The patient is initially
[11]. According to previous studies, this is due to comor- registered in SwEat at treatment onset and then followed
bidity with other psychiatric diagnoses [1, 7, 12], difficul- up once a year until EOT. Each patient might be initially
ties in responding to therapy when in starvation [13], registered more than once, since a patient is initially regis-
patients’ denial of their problems [1, 7] and unwillingness tered again if terminating treatment and later on entering a
to gain weight [14]. It has also been suggested that the new treatment episode. SwEat includes data from patients
physical and cognitive development that occurs during of all ages and both genders. A total of 17611 initial regis-
adolescence, in addition to major life changes like moving trations were made in SwEat between 1999 and 2014,
away from home, make treatment planning complex when there was a change in methodology and the original
[1, 2]. Approximately 20–40% relapse within the first version of SwEat was revised.
year after end of treatment (EOT) [15, 16], a rate that
is somewhat lower among adolescents than among Study sample
adults [16]. Complete recovery is expected in about All patients who met the following criteria were in-
50% of AN cases [8, 17, 18]. cluded in the study: medical or self-referral to one of the
Only a few naturalistic studies have been conducted participating treatment units between 1999 and 2014,
within the field of adolescent ED, (e.g. [19, 20]). 13–19 years of age at initial entry into SwEat and
Lindstedt et al. Journal of Eating Disorders (2017) 5:4 Page 3 of 10

diagnosed with AN or subthreshold AN according to included mainly information about patients who entered
DSM-IV [21], which during the years examined consti- treatment for the first time. Excluded were patients who i)
tuted the basis for diagnoses at Swedish ED units (see were followed up but had an incomplete follow-up regis-
Fig. 1 for a flow chart). The patients were diagnosed by tration, due for instance to inaccurate data, or ii) were still
experienced staff in consultation with multidisciplinary in treatment when the data collection was discontinued in
teams, and at most units on the basis of a structured 2014.
interview guide. Since 2008 the Structured Eating Dis- In the first analysis, when examining sample characteris-
order Interview (SEDI) has been used at Swedish ED tics at treatment onset, all patients remaining considering
units [27] and before that the most commonly used the mentioned criteria above were included. In the second
interview guide was the Rating of Anorexia and Bulimia and third analysis we included only patients who were
interview (RAB) [28]. In the present study, we focused registered at EOT as “in remission” or as completed for
on individuals instead of treatments and therefore other reasons. In the second analysis, these patients were
included solely information about the first treatment divided into three groups based on different time periods
episode for patients who had more than one episode for treatment onset; Period 1 (1999–2004), Period 2
registered. By choosing the first treatment episode, we (2005–2009) and Period 3 (2010–2014). A total of 83

Fig. 1 Flowchart, SwEat registrations of patients 13–19 years during 1999–2014

Lindstedt et al. Journal of Eating Disorders (2017) 5:4 Page 4 of 10

treatment units were represented in the study, of which Table 1 Data collected at SwEat registration1
42 were specialist ED units. Patients in the present study Initial Follow-up
received inpatient, day patient and/or outpatient treat- registration registration
ment, such as individual psychotherapy, family therapy Does the patient have symptoms consistent X
and group interventions. with a specified or unspecified ED, according
to DSM-IV? (Yes/No)
Of the patients in the total sample (n = 3997), 35%
Does the clinic intend to treat the patient? X
were lost to follow-up in SwEat. Comparisons between (Yes/No)
followed up and non-followed up patients regarding
Has the patient been informed about SwEat X
baseline characteristics showed only a few differences and given his/her oral consent for registration?
between the two groups, of which the most obvious was (Yes/No)
that followed-up patients had more social complications Civic registration number (YYYY-MM-DD-XXXX, X X
at treatment onset (15.6% vs 9.9%, p = <.001). Further- the last four digits comprise the Swedish social
more, the patients lost to follow-up had been ill for security number and specify gender)
rather longer when entering treatment (2.2 years vs Date of treatment onset (YYYY-MM-DD) X
1.9 years, p = <.001) and were younger at first symptoms The patient’s current ED diagnosis (DSM-IV X X
(14.4 years vs 14.7 years, p = <.001). Axis I/No current ED)
The patient’s age at onset of ED symptoms X
(years)
Measures The patient’s current weight (kg, to one X X
SwEat requires information about the following variables decimal)
used in the present study (Table 1). The patient’s current height (cm, to one X X
The registration form contains boxes for each re- decimal)
sponse alternative, and the system requires that all Are there one or several factors that clearly X
boxes are ticked before the form can be submitted. complicate treatment? (Yes, of psychiatric
Even so, most of the variables have 1–3% missing or nature/Yes, of somatic nature/Yes, of social
nature/No)
invalid answers.
Who referred the patient to the unit? (Patient/ X
In the present study the following variables were se- Relative/Other treatment unit or school)
lected as outcome measures:
What previous contact with the health care X
services did the patient have for the eating
disorder? (This is the first contact/Previous
Remission contact of an occasional nature/Previous
Patients not fulfilling criteria for any ED diagnosis at treatment)
follow-up were categorized as being in remission. Is the patient living alone or with others? X X
(Single/With children/With parents/With
partner/Other)
Weight status The patient’s employment (Studying/Working/ X X
Height and weight, either measured by a therapist or On sick leave)
self-reported by the patient, were used to calculate Is the treatment finished? (Yes/No) X
the patients’ BMI at initial registration and follow-up.
If the treatment is finished: What date? X
Based on the BMI percentile method for calculating (YY-MM-DD)
expected body weight, we assessed the patients as If the treatment is finished: How did it end? X
being within a normal or low weight range. This was (In agreement between patient and therapist
done in accordance to a previous study describing /Patient terminated treatment prematurely/
Patient was referred to another treatment
and recommending this method [29] and by using unit/Other reason)
Swedish reference values for BMI, adjusted for age 1
This table only includes data presented in the study. The SwEat registration
and gender [30]. contains additional data

Premature termination of treatment Treatment duration

The term is used for patients who do not complete treat- Treatment duration was measured in months and pos-
ment, regardless of reason [31]. In the present study, since sible to calculate for the patients who had a registered
there were no prescribed treatment doses and the length date for their first and last treatment session (n = 1904).
of treatment was not determined at treatment onset, the
term was used to categorize patients terminating treat- Sick leave
ment either on their own or their parents’ initiative or due At both initial registration and follow-up, information
to referral to another treatment unit. on employment is requested and clinicians are asked to
Lindstedt et al. Journal of Eating Disorders (2017) 5:4 Page 5 of 10

specify if the patient is on sick leave from work or of the cases, of which most were psychiatric. In a few
school. In the present study, we did not differentiate cases patients were registered as being on sick leave.
between patients on full or part-time sick leave. Approximately 60% of the patients had been referred to
treatment by for instance another treatment unit or a
Statistical analyses school health service. On average the patients had been
Statistical analyses were carried out using IBM SPSS ill for two years when entering treatment. Including only
Statistics 22. In order to compare variable values be- those entering treatment for the first time (n = 2737)
tween two different patient groups (e.g. followed up and illness duration was approximately 1.7 years (min = 0,
non-followed up patients) we used Pearson’s chi-square max = 11.9; SD = 1.6).
test and independent samples t-test. When exploring Separate analyses of patients who were registered at
possible differences between three time periods we used EOT as “in remission” or as completed for other reasons,
Pearson’s Chi-square test and one-way ANOVA, and ana- showed that just over two thirds ended treatment ac-
lyzed post hoc by examining possible differences between cording to an initial treatment plan or because they were
two groups at one time and by using Scheffe’s post hoc in remission. Other patients ended treatment prema-
test. Finally, we conducted logistic and multiple logistic turely, either on their own or their parents’ initiative (n
regressions in order to examine factors associated with = 283, 12.9%) or due to referral to another treatment
remission as an outcome variable. The regressions were unit (n = 350, 15.9%). The average treatment duration
performed using only those independent variables found was approximately 15 months (min = 1, max = 135). Just
to differ significantly between the groups In remission and above 55% of the patients were in remission at EOT and
Not in remission. In order to correct for multiple analyses, 16% of the patients were still considered to have a low
we used Bonferroni correction with thresholds set at weight (Table 3). Separate analyses of patients who
p = <.001 throughout the study. ended according to plan (n = 1564) revealed that just
under 70% were in remission at EOT and approximately
Results 10% were considered to have a low weight.
Most patients were adolescent girls who, at the time of The number of patients who were considered to have
treatment onset, were studying and living at home with a low weight at treatment onset were lower in period 3
their parents or other relatives (Table 2). Approximately and for each period relatively fewer patients had been
60% were considered to have a low weight when enter- entering treatment with an AN diagnosis. The propor-
ing treatment and almost as many had an AN diagnosis. tion of patients entering treatment with experiences of
One third of the patients had previous experiences of previous ED treatment and social or psychiatric compli-
treatment for ED and complicating social, psychiatric or cations were lower in period 2 and 3. Examination of
somatic factors were registered in more than one third treatment outcome revealed that treatment duration had
shortened for each time period. There were also a
Table 2 Total sample characteristics at treatment onset reduced number of patients who were considered to
Total sample have a low weight at EOT in period 3, and a larger number
(n = 3997) of patients in remission.
Girls (%) 3823 (95.6) Patients who terminated treatment prematurely had a
Studying (%) 3574 (89.4) decreased chance of achieving remission (Table 4). Remis-
Living with parents/other relatives (%) 3785 (94.7) sion was more likely for patients who entered treatment
in time period 3 compared to period 1. Also, although not
Low weight (%) 2385 (59.7)
significant at < .001, there was a clear tendency in the
AN (%) 2284 (57.1)
adjusted analyses that longer treatment duration was
Previous ED treatment (%) 1221 (30.5) positively correlated to remission.
Social complications (%) 543 (13.6)
Psychiatric complications (%) 685 (17.1) Discussion
Somatic complications (%) 149 (3.7) This study is one of few that investigates treatment out-
come in a naturalistic setting for adolescents with full or
Sick leave (%) 265 (6.6)
subthreshold AN. The main results show that 55% of the
Referred to treatment by other 2424 (60.6)
treatment unit or school (%)
participants were rated by clinicians as being in remission
and approximately 85% were within a healthy weight range
Age at first symptoms of ED M (SD) 14.6 (1.9)
at EOT. These results are in line with other naturalistic
Age when entering treatment M (SD) 16.6 (1.8) studies examining treatment outcome among adolescents
Illness duration at treatment onset (in years) M (SD) 2.0 (1.8) with AN (e.g. [19, 20, 32]). However, some of these studies
M mean, SD standard deviation are old and/or differ from our study in elementary aspects,
Lindstedt et al. Journal of Eating Disorders (2017) 5:4 Page 6 of 10

Table 3 Treatment characteristics and treatment outcome among patients with completed treatments; results for total sample and
comparisons between three time periods
Completed Time period Sign. Post hoc
treatments
(n = 2195) 1 1999–2004 2 2005–2009 3 2010–2014
(n = 457) (n = 1219) (n = 519)
Baseline characteristics
Age at first ED symptoms M (SD) 14.7 (1.9) 14.6 (1.9) 14.8 (1.8) 14.5 (1.9) .007 1-2 .250
2–3 .007
1–3 .499
Age at treatment onset M (SD) 16.6 (1.8) 16.5 (1.9) 16.6 (1.8) 16.4 (1.8) .071 1-2 .644
2–3 .058
1–3 .511
Referred to treatment by other 1374 (62.6) 300 (65.6) 771 (63.2) 303 (58.4) .050 1-2 .363
treatment unit or school (%)
2–3 .056
1–3 .020
Previous ED treatment (%) 667 (30.4) 199 (43.5) 355 (29.1) 113 (21.8) <.001 1-2 <.001
2–3 .002
1–3 <.001
Social complications (%) 355 (16.2) 103 (22.6) 183 (15.0) 69 (13.3) <.001 1-2 <.001
2–3 .352
1–3 <.001
Psychiatric complications (%) 362 (16.5) 104 (22.9) 189 (15.5) 69 (13.3) <.001 1-2 <.001
2–3 .236
1–3 <.001
Somatic complications (%) 88 (4.0) 23 (5.1) 45 (3.7) 20 (3.9) .440 1-2 .209
2–3 .871
1–3 .362
AN (%) 1240 (56.5) 310 (67.8) 683 (56.0) 247 (47.6) <.001 1-2 <.001
2–3 .001
1–3 <.001
Low weight (%) 1231 (56.1) 281 (61.5) 697 (57.2) 253 (48.7) <.001 1-2 .111
2–3 .001
1–3 <.001
Sick leave (%) 137 (6.2) 44 (9.6) 74 (6.1) 19 (3.7) .001 1-2 .011
2–3 .041
1–3 <.001
Outcome variables
Remission (%) 1220 (55.6) 221 (48.4) 682 (55.9) 317 (61.1) <.001 1-2 .006
2–3 .048
1–3 <.001
Low weight (%) 358 (16.3) 97 (21.3) 196 (16.1) 65 (12.5) .001 1-2 .012
2–3 .057
1–3 <.001
Premature termination of treatment (%) 633 (28.8) 127 (27.8) 357 (29.3) 149 (28.7) .832 1-2 .547
2–3 .808
1–3 .750
Treatment duration (months) M (SD)* 14.8 (11.8) 19.2 (15.8) 14.4 (10.8) 11.4 (7.7) <.001 1-2 <.001
2–3 <.001
1–3 <.001
Sick leave (%) 72 (3.3) 22 (4.8) 44 (3.6) 6 (1.2) .004 1-2 .248
2–3 .005
1–3 .001
*(n = 1904)
M mean, SD standard deviation
Lindstedt et al. Journal of Eating Disorders (2017) 5:4 Page 7 of 10

Table 4 Logistic and multiple logistic regression analyses with remission as an outcome variable
In remission Not in Unadjusted Adjusted
(n = 1220) remission
β (95% CI) Sign. β (95% CI) Sign.
(n = 975)
Previous ED treatment (%) Yes 363 (29.8) 304 (31.2) 0.94 (0.78 1.12) .471 1.02 (0.81 1.28) .873
No 857 (70.2) 671 (68.8) Ref
Psychiatric complications Yes 174 (14.3) 188 (19.3) 0.70 (0.56 0.88) .002 0.84 (0.63 1.10) .204
(onset) (%)
No 1043 (85.7) 787 (80.7) Ref.
AN (onset) (%) Yes 688 (56.4) 552 (56.6) 0.99 (0.84 1.17) .917 1.05 (0.82 1.34) .727
No 532 (43.6) 423 (43.4) Ref.
Low weight (onset) (%) Yes 670 (54.9) 561 (57.5) 0.90 (0.76 1.07) .219 0.79 (0.62 1.00) .054
No 550 (45.1) 414 (42.5) Ref.
Treatment duration 15.4 (10.9) 14.1 (12.6) 1.01 (1.00 1.02) .013 1.02 (1.01 1.02) .001
(months) M (SD)
Premature termination of treatment (%) Yes 128 (10.5) 503 (51.6) 0.11 (0.09 0.14) <.001 0.15 (0.12 0.19) <.001
No 1092 (89.5) 472 (48.4) Ref.
Time periods (%) 1 1999-2004 221 (18.1) 236 (24.2) 0.60 (0.46 0.77) <.001 0.51 (0.37 0.70) <.001
2 2005-2009 682 (55.9) 537 (55.1) 0.81 (0.66 1.00) .048 0.87 (0.68 1.12) .279
3 2010-2014 317 (26.0) 202 (20.7) Ref.
M mean, SD standard deviation

including assessment intervals and outcome estimates. It difficult. Despite a BMI within a seemingly normal range,
has been suggested that the definition of recovery used in a a young patient may have a low weight or even be
study has an important impact on outcome estimates [33]. underweight in relation to his or her own weight curve.
In the present study, we refrained from defining recovery This information is unfortunately hard to capture within
and confined ourselves to only use the term “in remission” SwEat, because of the large number of patients. A normal
for patients not fulfilling criteria for any ED diagnosis. It is weight does not necessarily mean that the patient is
not possible either to make a full comparison of our results healthy or recovered, and suffering from AN can be crit-
with randomized controlled trials within this field (e.g. [15, ical regardless of weight [2]. However, it is suggested in
34–36]), due for instance to divergent study structures and previous studies that BMI is an important prognostic
different ways of measuring outcome. factor [37] and that significant weight gain at EOT is a reli-
Approximately 70% of those who ended treatment able predictor of recovery in adolescents with AN [38, 39].
according to plan were rated as being in remission and Almost 60% of the adolescents had an AN diagnosis at
90% were within a healthy weight range. This indicates treatment onset, which in comparison to what is pre-
the importance of completing treatment, which will be sented in previous studies is a fairly large proportion
discussed later on. [22, 40]. In these studies it is suggested that the majority
The results in the present study also show that remis- of adolescents seeking ED treatment have variants of
sion rates and weight recovery increased over time, while subthreshold diagnoses. Considering treatment outcome,
treatment duration decreased. The fact that patients over the results in our study did not show any differences be-
the years became healthier when entering treatment is a tween patients with AN or subthreshold AN. In fact, the
possible explanation. However, this cannot fully explain number of patients in remission did not differ at all
these results since neither low weight nor complicating between the two groups. This corresponds to previous
factors at treatment onset was associated to a poor out- results suggesting that patients with AN or subthreshold
come. The results might therefore indicate that treatment AN in general suffer from symptoms to the same extent
has become more effective over the past 15 years. This [22–24], but runs counter to another study suggesting
seems promising, but needs to be studied further since that recovery is eight times more likely among patients
there is not enough knowledge about causal factors and with subthreshold AN [41].
the generalizability of such a trend. Patients in time period 2 and 3 were more often
The large number of patients within a healthy weight considered to have a normal weight when entering
range is, needless to say, a positive result. However, when treatment, were less often diagnosed with AN, had
studying adolescents it is important to bear in mind that a fewer experiences of previous ED treatment and less
categorization in low and normal weight based on BMI is social and psychiatric complicating factors. However,
Lindstedt et al. Journal of Eating Disorders (2017) 5:4 Page 8 of 10

there was no difference in illness duration at treat- patients [4, 12, 41, 46]. For adolescents, psychiatric co-
ment onset between the different time periods, as one morbidity comprises mainly mood- and anxiety disor-
could expect. Instead, adolescents during later years ders, obsessive-compulsive disorder, substance abuse
might have been seeking treatment for less serious and personality disorders [2, 4]. The results in the
conditions, perhaps due to easier access to health care present study indicate that psychiatric complications
and increased awareness of ED in society. might be associated with a poor outcome, which also
In our study, the average age when entering treatment is in line with results from previous studies [46, 47].
was 16.6 years, which is in accordance to results from a For example, Wentz and colleagues found that psychiatric
British study suggesting that the peak age of presentation complications might affect vulnerability for AN as well as
for treatment is 15–19 years [5]. Age at first symptoms of treatment outcome [46].
illness was on average 14.6 years, which indicates approxi- There are some limitations to this study. Considerable
mately two years of illness duration at treatment onset. attrition at follow-up in SwEat is one, over which we
When excluding patients with experiences of previous ED unfortunately had no control when designing the study.
treatment, illness duration at treatment onset was slightly The amount of follow-up registrations in SwEat varies
shorter. In previous studies it is suggested that duration greatly between different units, probably due to varying
between onset of illness and initiation of treatment is follow-up procedures. In general, approximately 60% of
often rather long, in particular when it comes to those initial registrations were lost to follow-up one year later
with an early onset of illness [7, 9]. This may partly reflect [48], which might have to do with the fact that young
the fact that many people with AN do not see their symp- people often wish to terminate their treatment quickly.
toms as problematic but more as a part of their identity The loss of patients at follow-up affects the generalizability
and that they lack internal motivation to recover [1, 7]. of the results, although we did not find any differences of
Many adolescents are likely to have atypical presentations clinical relevance between followed-up and non-followed
of ED, which increases the risk for delayed diagnoses and up patients in the present study. Another limitation is the
significant complications [1, 2, 42, 43]. Approximately one missing data considering some of the variables in SwEat,
third of the patients in the present study terminated treat- due for instance to different technical issues or errors when
ment prematurely, either on their own or their parents’ registrations were made. As mentioned earlier, approxi-
initiative or due to referral to another treatment unit. This mately 1–3% of the answers throughout the register are
corresponds to results from previous studies, suggesting a missing or incorrect. The fact that height and weight in
proportion of 20–40% [11]. Premature termination of some cases were self-reported by the patient might be
treatment is considered a problem within several psychi- considered a limitation, although previous results suggest
atric disorders and in particular within the field of ED and that self-reported height and weight are reliable [49]. Some
AN [44, 45]. For example, as this study also showed, major limitations to this study are that we had to judge the
terminating treatment prematurely reduces the chance of reliability of what clinicians have reported for some of the
achieving remission while completing treatment increases variables and that we, due to the many years examined and
the chances of a good outcome [11, 35]. As mentioned the large amount of participating units, lacked control over
earlier, as many as 70% of those who ended treatment the assessments of symptoms and diagnoses. No inter-rater
according to plan were in remission and 90% within a agreement estimates were made and the procedure for es-
healthy weight range. Which clinical characteristics and tablishing ED diagnoses varied over time as well as between
factors that can be associated with premature termination units in different parts of the country. Also, the fact that
have yet to be discovered, but one suggestion is dis- some of the variables (e.g. age at first symptoms) were
crepancy between patient preferences and expectations assessed retrospectively may have led to memory bias.
about treatment that may account for non-adherence These limitations, in addition to the fact that data might be
[1]. It may also be linked with treatment dissatisfac- affected by selection bias, are related to the naturalistic
tion, which will be explored in an upcoming study design of the study and mentioned also in previous studies
based on data from SwEat. In the present study, treat- as disadvantages with naturalistic register studies [12, 50].
ment duration was approximately 15 months and it However, the design of the study also provides several
has been suggested that treatment should last at least strengths. The large-scale naturalistic setting secures the
six months for a desirable outcome [7, 20]. generalizability to a clinical environment and offers a
The fact that complicating social, psychiatric or somatic comparison for outcome data from treatment trials [20].
factors were registered in more than one third of the The naturalistic setting also provides a natural treatment
cases, of which most were psychiatric, is not surprising. environment for patients and clinicians, when daily
Although only a few studies have looked at social or routines can be followed despite study participation. An
somatic factors (e.g. [46]), psychiatric comorbidity is additional strength with the present study is the number of
well known to be a complicating factor for these participating units, providing good national coverage.
Lindstedt et al. Journal of Eating Disorders (2017) 5:4 Page 9 of 10

Future research would benefit from different forms of Competing interests

studies focusing on how different treatment settings and The authors declare that they have no competing interests.

approaches affect treatment outcome [51]. In recent

Consent for publication
years several different methods have been examined and
Not applicable.
tested, but nevertheless there are only a few uniform
recommendations that can be applied to patients of Ethics approval and consent to participate
different ages, with different diagnoses and in different Participants in the present study were identified through the Swedish
social situations. The most effective treatments for national quality register for eating disorder treatment (SwEat). Information is
registered in SwEat when it is established that the patient has given her/his
adolescents with AN include family based therapy [52], consent to registration. The study was approved by the Research Ethics
but it is important to take several aspects into consideration Committee of Uppsala (Dnr 2011/478).
when choosing a treatment model because of the hetero-
Received: 17 August 2016 Accepted: 26 January 2017
geneity of the patient group [53]. In the present study, no
distinction was made between specialist and non-specialist
ED units, but it has earlier been recommended that more References
resources should be devoted to specialist outpatient 1. Treasure JL. Applying evidence-based management to anorexia nervosa.
ED services with direct access from primary care for Postgrad Med J. 2016;0:1-7.
2. Campbell K, Peebles R. Eating disorders in children and adolescents: state of
better outcome [54]. the Art review. Pediatrics. 2014;134:582–92.
3. Smink FRE, van Hoeken D, Hoek HW. Epidemiology of eating disorders:
incidence, prevalence and mortality rates. Curr Psychiatry Rep. 2012;14:406–14.
Conclusions 4. Herpertz-Dahlmann B. Adolescent eating disorders: update on definitions,
Only a few naturalistic studies have focused on outcome symptomatology, epidemiology, and comorbidity. Child Adolesc Psychiatr
for patients in routine care. This study shows that approxi- Clin N Am. 2015;24:177–96.
5. Micali N, Hagberg KW, Petersen I, Treasure JL. The incidence of eating
mately 55% were rated by clinicians as being in remission disorders in the UK in 2000-2009: findings from the general practice
and approximately 85% were rated as being within a research database. BMJ Open. 2013;3:5.
healthy weight range at EOT. The results indicate that 6. Bailey AP, Parker AG, Colautti LA, Hart LM, Liu P, Hetrick SE. Mapping the
evidence for the prevention and treatment of eating disorders in young
treatment for adolescents with ED in Sweden has become people. J Eat Disord. 2014;2:5.
more effective over the past 15 years, with more patients 7. Neubauer K, Weigel A, Daubmann A, Wendt H, Rossi M, Lowe B, et al. Paths
reaching remission and a healthy weight after a shorter to first treatment and duration of untreated illness in anorexia nervosa: are
there differences according to age of onset? Eur Eat Disord Rev. 2014;22:292–8.
treatment duration. The results of the present study 8. Steinhausen H-C. Outcome of eating disorders. Child Adolesc Psychiatr Clin N Am.
contribute to the scope of treatment research, but 2009;18:225–42.
more research is needed into different forms of evi- 9. Franko DL, Keshaviah A, Eddy KT, Krishna M, Davis MC, Keel PK, et al. A
longitudinal investigation of mortality in anorexia nervosa and bulimia
dence, new research strategies and diversity of treat- nervosa. Am J Psychiatry. 2013;170:917–25.
ment approaches. 10. Zipfel S, Giel KE, Bulik CM, Hay P, Schmidt U. Anorexia nervosa: aetiology,
assessment, and treatment. Lancet Psychiatry. 2015;2:1099–111.
Abbreviations 11. Dejong H, Broadbent H, Schmidt U. A systematic review of dropout from
AN: Anorexia nervosa; BMI: Body mass index; DSM-IV: Diagnostic and treatment in outpatients with anorexia nervosa. Int J Eat Disord. 2012;45:635–47.
statistical manual of mental disorders IV; ED: Eating disorder; EOT: End of 12. Ulfvebrand S, Birgegard A, Norring C, Hogdahl L, von Hausswolff-Juhlin Y.
treatment; SwEat: Swedish national quality register for eating disorder Psychiatric comorbidity in women and men with eating disorders results
treatment from a large clinical database. Psychiatry Res. 2015;230:294–99.
13. Treasure J, Schmidt U. The cognitive-interpersonal maintenance model of
Acknowledgements anorexia nervosa revisited: a summary of the evidence for cognitive,
We would like to acknowledge the Swedish national quality register for socio-emotional and interpersonal predisposing and perpetuating factors.
eating disorder treatment (SwEat). J Eat Disord. 2013;1:13.
14. Williams S, Reid M. Understanding the experience of ambivalence in anorexia
nervosa: the maintainer’s perspective. Psychol Health. 2010;25:551–67.
Funding
15. Herpertz-Dahlmann B, Schwarte R, Krei M, Egberts K, Warnke A, Wewetzer C,
Funding support for this study was provided by Region Örebro County and
et al. Day-patient treatment after short inpatient care versus continued
Örebro University.
inpatient treatment in adolescents with anorexia nervosa (ANDI): a multicentre,
randomised, open-label, non-inferiority trial. Lancet. 2014;383:122–29.
Availability of data and materials 16. Carter JC, Mercer-Lynn KB, Norwood SJ, Bewell-Weiss CV, Crosby RD, Woodside DB,
Data from the current study are not publicly available in order to maintain et al. A prospective study of predictors of relapse in anorexia nervosa: implications
participant confidentiality, but are available from the corresponding author for relapse prevention. Psychiatry Res. 2012;200:518–23.
on reasonable request with approval from the relevant institutional review 17. Lock J. Evaluation of family treatment models for eating disorders. Curr Opin
board. Psychiatry. 2011;24:274–79.
18. Harrington BC, Jimerson M, Haxton C, Jimerson DC. Initial evaluation,
Authors’ contributions diagnosis, and treatment of anorexia nervosa and bulimia nervosa.
KL, LK and SAG contributed to the conception and design of the study. Am Fam Physician. 2015;91:46–52.
KL and SAG contributed to the acquisition and preparation of the data. 19. Rosling A, Salonen Ros H, Swenne I. One-year outcome and incidence of
KL conducted data analyses and KL, LK and SAG made substantial contributions anorexia nervosa and restrictive eating disorders among adolescent girls
to the interpretation of data. KL drafted the manuscript while LK and SAG treated as out-patients in a family-based setting. Ups J Med Sci. 2016;121:50–9.
reviewed it critically for accuracy and theoretical integrity and contributed to 20. Thompson-Brenner H, Boisseau CL, Satir DA. Adolescent eating disorders:
the writing. KL, LK and SAG read and approved the final manuscript. treatment and response in a naturalistic study. J Clin Psychol. 2010;66:277–301.
Lindstedt et al. Journal of Eating Disorders (2017) 5:4 Page 10 of 10

21. American Psychiatric Association. Diagnostic and statistical manual of mental 47. Vall E, Wade TD. Predictors of treatment outcome in individuals with
disorders. 4th ed. Washington, DC: American Psychiatric Association; 2000. eating disorders: a systematic review and meta-analysis. Int J Eat Disord.
22. Eddy KT, Celio Doyle A, Hoste RR, Herzog DB, le Grange D. Eating disorder 2015;48:946–71.
not otherwise specified in adolescents. J Am Acad Child Adolesc Psychiatry. 48. Gustafsson SA, Norring C, Norring S. Riksät - nationellt kvalitetsregister för
2008;47:156–64. ätstörningsbehandling. In: Årsrapport 2013. Örebro: Region Örebro län; 2014.
23. Peebles R, Hardy KK, Wilson JL, Lock JD. Are diagnostic criteria for eating 49. Swenne I, Belfrage E, Thurfjell B, Engstrom I. Accuracy of reported weight
disorders markers of medical severity? Pediatrics. 2010;125:e1193–201. and menstrual status in teenage girls with eating disorders. Int J Eat Disord.
24. Le Grange D, Crosby RD, Engel SG, Cao L, Ndungu A, Crow SJ, et al. DSM-IV- 2005;38:375–9.
defined anorexia nervosa versus subthreshold anorexia nervosa (EDNOS-AN). 50. Runfola CD, Thornton LM, Pisetsky EM, Bulik CM, Birgegard A. Self-image
Eur Eat Disord Rev. 2013;21:1–7. and suicide in a Swedish national eating disorders clinical register. Compr
25. American Psychiatric Association. Diagnostic and statistical manual of mental Psychiatry. 2014;55:439–49.
disorders. 5th ed. Washington, DC: American Psychiatric Association; 2013. 51. Clinton D. Towards an ecology of eating disorders: creating sustainability through
26. Gustafsson SA, Norring C, Norring S. Riksät - nationellt kvalitetsregister för the integration of scientific research and clinical practice. Eur Eat Disord Rev.
ätstörningsbehandling. In: Årsrapport 2014. Örebro: Region Örebro län; 2015. 2010;18:1–9.
27. Birgegard A, Bjorck C, Clinton D. Quality assurance of specialised treatment 52. Brown TA, Keel PK. Current and emerging directions in the treatment of
of eating disorders using large-scale internet-based collection systems: eating disorders. Subst Abus. 2012;6:33–61.
methods, results and lessons learned from designing the stepwise database. 53. Lindstedt K, Neander K, Kjellin L, Gustafsson SA. Being me and being us -
Eur Eat Disord Rev. 2010;18:251–9. adolescents’ experiences of treatment for eating disorders. J Eating Disord.
28. Nevonen L, Broberg AG, Clinton D, Norring C. A measure for the assessment 2015;3:1–11.
of eating disorders: reliability and validity studies of the rating of anorexia 54. House J, Schmidt U, Craig M, Landau S, Simic M, Nicholls D, et al. Comparison
and bulimia interview - revised version (RAB-R). Scand J Psychol. 2003;44:303–10. of specialist and nonspecialist care pathways for adolescents with anorexia
29. Le Grange D, Doyle PM, Swanson SA, Ludwig K, Glunz C, Kreipe RE. nervosa and related eating disorders. Int J Eat Disord. 2012;45:949–56.
Calculation of expected body weight in adolescents with eating disorders.
Pediatrics. 2012;129:e438–46.
30. Karlberg J, Luo ZC, Albertsson-Wikland K. Body mass index reference values
(mean and SD) for Swedish children. Acta Paediatr. 2001;90:1427–34.
31. Sly R. What’s in a name? Classifying’the dropout’ from treatment for
anorexia nervosa. Eur Eat Disord Rev. 2009;17:405–7.
32. Steinhausen HC, Boyadjieva S, Griogoroiu-Serbanescu M, Neumarker KJ.
The outcome of adolescent eating disorders: findings from an international
collaborative study. Eur Child Adolesc Psychiatry. 2003;12 Suppl 1:91–8.
33. Bjork T, Clinton D, Norring C. The impact of different outcome measures on
estimates of remission in a 3-year follow-up of eating disorders. Eur Eat
Disord Rev. 2011;19:2–11.
34. Lock J, Le Grange D, Agras WS, Moye A, Bryson SW, Jo B. Randomized
clinical trial comparing family-based treatment with adolescent-focused
individual therapy for adolescents with anorexia nervosa. Arch Gen Psychiatry.
2010;67:1025–32.
35. Gowers SG, Clark A, Roberts C, Griffiths A, Edwards V, Bryan C, et al.
Clinical effectiveness of treatments for anorexia nervosa in adolescents:
randomised controlled trial. Br J Psychiatry. 2007;191:427–35.
36. Agras WS, Lock J, Brandt H, Bryson SW, Dodge E, Halmi KA, et al. Comparison
of 2 family therapies for adolescent anorexia nervosa: a randomized parallel
trial. JAMA Psychiatry. 2014;71:1279–86.
37. Rosling AM, Sparen P, Norring C, von Knorring AL. Mortality of eating disorders:
a follow-up study of treatment in a specialist unit 1974-2000. Int J Eat Disord.
2011;44:304–10.
38. Le Grange D, Accurso EC, Lock J, Agras S, Bryson SW. Early weight gain
predicts outcome in two treatments for adolescent anorexia nervosa.
Int J Eat Disord. 2014;47:124–9.
39. Treasure J, Russell G. The case for early intervention in anorexia nervosa:
theoretical exploration of maintaining factors. Br J Psychiatry. 2011;199:5–7.
40. Bravender T, Bryant-Waugh R, Herzog D, Katzman D, Kreipe RD, Lask B, et al.
Classification of child and adolescent eating disturbances. Workgroup for
classification of eating disorders in children and adolescents (WCEDCA).
Int J Eat Disord. 2007;40 Suppl 1:117–22.
41. Silén Y, Raevuori A, Jüriloo E, Tainio V-M, Marttunen M, Keski-Rahkonen A.
Typical versus atypical anorexia nervosa among adolescents: clinical Submit your next manuscript to BioMed Central
characteristics and implications for ICD-11. Eur Eat Disord Rev. 2015;23:345–51. and we will help you at every step:
42. Sim LA, Lebow J, Billings M. Eating disorders in adolescents with a history of
obesity. Pediatrics. 2013;132:e1026–30. • We accept pre-submission inquiries
43. Silber TJ. Anorexia nervosa in children and adolescents: diagnosis, treatment • Our selector tool helps you to find the most relevant journal
and the role of the pediatrician. Minerva Pediatr. 2013;65:1–17.
• We provide round the clock customer support
44. Jordan J, McIntosh VV, Carter FA, Joyce PR, Frampton CM, Luty SE, et al.
Clinical characteristics associated with premature termination from outpatient • Convenient online submission
psychotherapy for anorexia nervosa. Eur Eat Disord Rev. 2014;22:278–84. • Thorough peer review
45. Watson HJ, Bulik CM. Update on the treatment of anorexia nervosa: review
• Inclusion in PubMed and all major indexing services
of clinical trials, practice guidelines and emerging interventions. Psychol
Med. 2013;43:2477–500. • Maximum visibility for your research
46. Wentz E, Gillberg IC, Anckarsater H, Gillberg C, Rastam M. Adolescent-onset
anorexia nervosa: 18-year outcome. Br J Psychiatry. 2009;194:168–74. Submit your manuscript at
www.biomedcentral.com/submit