See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/247000186

Focal psychodynamic psychotherapy of

anorexia nervosa: A treatment manual

Article in Psychotherapeut · January 2009

CITATIONS READS

3 1,105

5 authors, including:

Henning Schauenburg Hans-Christoph Friederich

Universität Heidelberg, Universität Heidelberg
184 PUBLICATIONS 1,213 CITATIONS 121 PUBLICATIONS 1,933 CITATIONS

SEE PROFILE SEE PROFILE

Stephan Zipfel Wolfgang Herzog

University of Tuebingen Universität Heidelberg
458 PUBLICATIONS 5,615 CITATIONS 458 PUBLICATIONS 8,511 CITATIONS

SEE PROFILE SEE PROFILE

All in-text references underlined in blue are linked to publications on ResearchGate, Available from: Henning Schauenburg
letting you access and read them immediately. Retrieved on: 11 November 2016
 Articles

Focal psychodynamic therapy, cognitive behaviour therapy,

and optimised treatment as usual in outpatients with
anorexia nervosa (ANTOP study): randomised controlled trial
Stephan Zipfel, Beate Wild, Gaby Groß, Hans-Christoph Friederich, Martin Teufel, Dieter Schellberg, Katrin E Giel, Martina de Zwaan,
Andreas Dinkel, Stephan Herpertz, Markus Burgmer, Bernd Löwe, Sefik Tagay, Jörn von Wietersheim, Almut Zeeck, Carmen Schade-Brittinger,
Henning Schauenburg, Wolfgang Herzog on behalf of the ANTOP study group*

Summary
Background Psychotherapy is the treatment of choice for patients with anorexia nervosa, although evidence of efficacy Published Online
is weak. The Anorexia Nervosa Treatment of OutPatients (ANTOP) study aimed to assess the efficacy and safety of October 14, 2013
http://dx.doi.org/10.1016/
two manual-based outpatient treatments for anorexia nervosa—focal psychodynamic therapy and enhanced cognitive S0140-6736(13)61746-8
behaviour therapy—versus optimised treatment as usual.
See Online/Comment
http://dx.doi.org/10.1016/
Methods The ANTOP study is a multicentre, randomised controlled efficacy trial in adults with anorexia nervosa. S0140-6736(13)61940-6
We recruited patients from ten university hospitals in Germany. Participants were randomly allocated to 10 months of *See end of report for ANTOP
treatment with either focal psychodynamic therapy, enhanced cognitive behaviour therapy, or optimised treatment as study group members
usual (including outpatient psychotherapy and structured care from a family doctor). The primary outcome was Department of Psychosomatic
weight gain, measured as increased body-mass index (BMI) at the end of treatment. A key secondary outcome Medicine and Psychotherapy,
University Hospital Tübingen,
was rate of recovery (based on a combination of weight gain and eating disorder-specific psycho­ pathology). Tübingen, Germany
Analysis was by intention to treat. This trial is registered at http://isrctn.org, number ISRCTN72809357. (Prof S Zipfel MD, G Groß PhD,
M Teufel MD, K E Giel PhD);
Findings Of 727 adults screened for inclusion, 242 underwent randomisation: 80 to focal psychodynamic therapy, Center for Psychosocial
Medicine, Department of
80 to enhanced cognitive behaviour therapy, and 82 to optimised treatment as usual. At the end of treatment, 54 patients General Internal Medicine and
(22%) were lost to follow-up, and at 12-month follow-up a total of 73 (30%) had dropped out. At the end of treatment, Psychosomatics, Heidelberg
BMI had increased in all study groups (focal psychodynamic therapy 0·73 kg/m², enhanced cognitive behaviour University Hospital,
Heidelberg, Germany
therapy 0·93 kg/m², optimised treatment as usual 0·69 kg/m²); no differences were noted between groups
(B Wild PhD, H-C Friederich MD,
(mean difference between focal psychodynamic therapy and enhanced cognitive behaviour therapy –0·45, D Schellberg PhD,
95% CI –0·96 to 0·07; focal psychodynamic therapy vs optimised treatment as usual –0·14, –0·68 to 0·39; enhanced Prof H Schauenburg MD,
cognitive behaviour therapy vs optimised treatment as usual –0·30, –0·22 to 0·83). At 12-month follow-up, the mean Prof W Herzog MD);
Department of Psychosomatic
gain in BMI had risen further (1·64 kg/m², 1·30 kg/m², and 1·22 kg/m², respectively), but no differences between
Medicine and Psychotherapy,
groups were recorded (0·10, –0·56 to 0·76; 0·25, –0·45 to 0·95; 0·15, –0·54 to 0·83, respectively). No serious adverse University Hospital Erlangen,
events attributable to weight loss or trial participation were recorded. Erlangen, Germany
(Prof M de Zwaan MD); Clinic for
Psychosomatic Medicine and
Interpretation Optimised treatment as usual, combining psychotherapy and structured care from a family doctor,
Psychotherapy, University of
should be regarded as solid baseline treatment for adult outpatients with anorexia nervosa. Focal psychodynamic Technology Munich, Munich,
therapy proved ad­vantageous in terms of recovery at 12-month follow-up, and enhanced cognitive behaviour therapy Germany (A Dinkel PhD); Clinic
was more effective with respect to speed of weight gain and improvements in eating disorder psychopathology. for Psychosomatic Medicine
and Psychotherapy, LWL
Long-term outcome data will be helpful to further adapt and improve these novel manual-based treatment approaches.
University Hospital of the Ruhr,
University of Bochum,
Funding German Federal Ministry of Education and Research (Bundesministerium für Bildung und Forschung, Bochum, Germany
BMBF), German Eating Disorders Diagnostic and Treatment Network (EDNET). (Prof S Herpertz MD); Clinic for
Psychosomatic Medicine and
Psychotherapy, University
Introduction severity, poor prognosis, and low prevalence of the Hospital Münster, Münster,
Anorexia nervosa is associated with serious medical disorder are reasons why large randomised controlled Germany (Prof M Burgmer MD);
morbidity1,2 and pronounced psychosocial comorbidity.3 trials are needed and why difficulties arise in imple­ Institute and Outpatient Clinic
for Psychosomatic Medicine
It has the highest mortality rate of all mental disorders4,5 mentation of treatment studies.10 and Psychotherapy, University
and relapse happens frequently.6 The course of illness is According to international treatment guidelines, psycho­ Hospital Hamburg-Eppendorf,
very often chronic, particularly if left untreated.7 Partial therapy is the treatment of choice for patients with anor­ Hamburg, Germany
syndromes are also associated with adverse health exia, although no evidence clearly supports the efficacy of (Prof B Löwe MD); Clinic for
Psychosomatic Medicine and
outcomes. Quality of life for patients is poor, and the cost any specific form of psychotherapy.11 Guidelines from the Psychotherapy, LVR Hospital
and burden placed on individuals, families,1 and society UK’s National Institute for Health and Care Excellence Essen, University of
is high.8 The overall incidence of anorexia nervosa is at (NICE) outline 75 recommendations for the treatment of Duisburg-Essen, Essen,
least eight people per 100 000 per year, with an average anorexia nervosa.12 74 of these treatments have received a Germany (S Tagay PhD);
Department of Psychosomatic
prevalence of 0·3% in girls and young women.9 The grade of C, meaning that good quality, directly applicable

www.thelancet.com Published online October 14, 2013 http://dx.doi.org/10.1016/S0140-6736(13)61746-8 1

 Articles

Medicine and Psychotherapy, clinical studies are absent and that recommendations are and statistical manual of mental disorders, 4th edition
University Hospital of Ulm, based solely on the opinions, clinical experience, or both (DSM-IV); and a body-mass index (BMI) of 15–18·5 kg/m².
Ulm, Germany
(Prof J von Wietersheim PhD);
of respected authorities in the field. According to NICE Exclusion criteria were: current sub­stance abuse; use of
Department of Psychosomatic guidelines, psychological treatment of anorexia nervosa neuroleptic drugs; psychotic or bipolar disorder; serious
Medicine and Psychotherapy, aims to lessen risk, encourage weight gain and healthy unstable medical problems; and ongoing psychotherapy.
University Hospital Freiburg, eating, reduce other symptoms related to the eating We medically assessed patients at baseline and did a
Freiburg, Germany
(Prof A Zeeck MD); and
disorder, and facilitate psychological and physical recovery. comprehensive diagnostic assessment, which included
Coordination Center for Clinical In a Cochrane review of outpatient treatment for anorexia measuring weight and height and undertaking structured
Trials (KKS), Marburg, Germany nervosa,13 only seven small trials were identified, two of diagnostic interviews specific to psychiatry and eating
(C Schade-Brittinger MA) which included children or adolescents. Findings of two disorders. We obtained written informed consent from
Correspondence to: of the trials implied that treatment as usual might be less every participant at the baseline visit. Independent
Prof Stephan Zipfel, Department
of Psychosomatic Medicine and
effective than a specific psychotherapy. No particular research ethics committees at every participating centre
Psychotherapy, University treatment, however, was consistently superior to any approved the study.
Hospital Tübingen, Tübingen, other approach.
Germany In adults with anorexia nervosa, some evidence shows Randomisation and masking
stephan.zipfel@
med.uni-tuebingen.de
the effectiveness of outpatient focal psychodynamic The independent coordination centre for clinical trials
therapy and cognitive behaviour therapy.14–16 In one trial,17 (Marburg, Germany) did centralised randomisation.
at the end of the treatment period, a supportive therapy Patients were randomly assigned to 10 months of treat­
delivered by specialists was superior to two specific ment with either focal psychodynamic therapy, enhanced
psychotherapies, with respect to a combined global cognitive behaviour therapy, or optimised treatment as
outcome measure. However, long-term follow-up of this usual in a 1:1:1 ratio. We used the Rosenberg and Lachin
trial showed that interpersonal therapy was the most covariate-adaptive randomisation procedure23 based on
successful treatment.18 Findings of intervention studies Nordle and Brantmark’s design.24 This pro­cedure com­
applying deep-brain stimulation19 or adapting psycho­ bines elements of the minimisation approach (to
therapeutic approaches for patients with chronic anorexia optimally allocate a treatment to a particular patient
nervosa20 have also showed some promising results for based on his or her prognostic factor combination) with
this cohort. the biased-coin technique to avoid a deterministic treat­
Evidence accumulated thus far does not support any one ment allocation. We stratified randomisation by centre
particular psychotherapeutic method for the treat­ment of and duration of anorexia nervosa (≤6 years vs >6 years).
adults with anorexia nervosa.1,13 However, therapeutic sup­ After patients were randomised into groups, the indepen­
port from a non-specialist clinician might be less success­ dent centre faxed trial sites the treatment allocation.
ful than a specific form of psychotherapy provided by a Complete masking of participants was not feasible
specialist. Additionally, no evidence strongly advocates because a third of patients were allocated optimised treat­
drug treatment either in the acute or maintenance phase ment as usual and, therefore, were not treated at the
of the illness.21 Large, well designed psychotherapeutic respective centres. More information about the masking
See Online for appendix trials are needed urgently. We designed the Anorexia procedure is provided in the appendix (p 2).
Nervosa Treatment of OutPatients (ANTOP) study to
investigate the efficacy of two manual-based, psycho­ Procedures
therapeutic, eating disorder-specific out­patient therapies Patients allocated to either focal psychodynamic therapy
for adults with anorexia nervosa—focal psychodynamic or enhanced cognitive behaviour therapy received an
therapy and enhanced cognitive behaviour therapy— individual outpatient intervention based on standardised
compared with optimised treatment as usual. treatment manuals.25,26 To avoid contamination between
treatment arms, the two approaches were provided by
Methods different therapists, who were all skilled at the underlying
Study design and participants therapeutic approach (panel 1). Therapists received initial
ANTOP was a multicentre, randomised controlled efficacy 2-day training from experts (focal psychodynamic therapy:
trial in adult patients with anorexia nervosa. The trial WH, HS, H-CF; enhanced cognitive behaviour therapy:
protocol, outlining details on study design, has been C Fairburn), followed by annual training updates (focal
published elsewhere.22 Over a 2-year period, we screened psychodynamic therapy: WH, HS, H-CF; enhanced cog­
patients from outpatient departments of ten German nitive behaviour therapy: GG, MdZ). At every fourth
university departments of psychosomatic medicine and session, experienced local supervisors oversaw the thera­
psychotherapy (Bochum, Erlangen, Essen, Freiburg, pists’ work. Panel 1 outlines the essentials of the three
Hamburg, Heidelberg, Munich, Münster, Tübingen, and treatment manuals. Information about adherence control
Ulm) for inclusion in the study. Inclusion criteria were: and treatment fidelity is provided in the appendix (p 4).
adult patient (aged ≥18 years); female sex; a diagnosis of We did the study according to International Conference
anorexia nervosa or subsyndromal anorexia nervosa (one on Harmonisation Good Clinical Practice guidelines. We
diagnostic criterion absent), according to the diagnostic incorporated quality-control methods including case-report

2 www.thelancet.com Published online October 14, 2013 http://dx.doi.org/10.1016/S0140-6736(13)61746-8

 Articles

forms, independent data management, on-site monitoring, report to the independent data safety and monitoring
and documentation of adverse and severe adverse events. committee during annual meetings. After the study
Data management included regular checks for consistency protocol was published and finally approved by the
and plausibility, and queries if inconsistencies or missing independent coordination centre, two statisticians not
data were noted. In addition to an initiation site visit in involved in treatment and diagnostics of the ANTOP study
2007 and a close-out visit in 2011, the independent did biometric analyses.
coordination centre made three annual on-site data
monitoring visits to every study centre, according to Outcome measures
existing standard operating procedures. The main aim of We took measurements at five timepoints: at a baseline
the monitoring procedure was to verify patients’ safety and assessment (before randomisation); 4 months after treat­
the complete­ness, accuracy, and validity of the trial data, ment was started; 10 months after the start of treatment
and to comply with the study protocol. Additionally, the (which accorded with the end of treatment); after
principal investigators (SZ, WH, BW, and GG) had to 3 months of follow-up (short-term); and at a 12-month

Panel 1: Treatment provided

Overview restructuring, mood regulation, social skills, shape concern, and
We developed a framework of medical care for all patients in self-esteem. The treatment plan represents an extension of the
the ANTOP study. This framework included at least five regular focused version of enhanced cognitive behaviour therapy,
sessions of specialist assessment at the patient’s specific study combined with elements of the broad version. It focuses on
centre. To avoid and reduce medical complications during the education of patients about being underweight and starvation
study period, we asked all patients to see their family doctor at and helps patients initiate and maintain regular eating and
least once a month. Every individual study centre gave patients’ healthy weight gain. Enhancement of self-efficacy and
family doctors written instructions on how to provide care in self-monitoring are crucial elements of the entire treatment
relation to this study. Treatment was provided face-to-face by process. Therefore, therapists selected additional practice and
doctors and psychologists specialising in anorexia nervosa and homework worksheets, written in German, which they gave
the respective assigned treatment method. Additionally, we patients at the end of every session. The patients received these
implemented a rigorous system of supervision, adherence homework assignments (next steps) to ensure the generalisation
control, and treatment fidelity (appendix p 3). and transfer of therapeutic changes to daily life. Further details of
the two interventions are described in the treatment manuals
Focal psychodynamic therapy
and additional published materials.
At the beginning of focal psychodynamic therapy (FPT), we
identified psychodynamic foci with a standardised, Optimised treatment as usual
operationalised, psychodynamic diagnostic interview. The Patients assigned to optimised treatment as usual received
psychodynamic treatment manual can be divided roughly into support in accessing therapy and were given a list of established
three treatment phases. The first phase focuses mainly on outpatient psychotherapists with experience in treating eating
therapeutic alliance, pro-anorectic behaviour and ego-syntonic disorders and who work in accordance with German general
beliefs (attitudes and behaviour viewed as acceptable), and psychotherapy guidelines. Patients’ family doctors had an
self-esteem. In the second phase of treatment, main focus is active role in treatment and monitoring. In the German
placed on relevant relationships and the association between health-care system, psychotherapy for patients with eating
interpersonal relationships and eating (anorectic) behaviour. disorders—in particular, those with anorexia nervosa—is usually
The pertinent aspects of the final phase are the transfer to covered by health insurance. To further optimise the treatment
everyday life, anticipation of treatment termination, and as usual approach, patients’ family doctors had three roles.
parting. Before every treatment session, an independent First, they were asked to take regular weight measurements, do
assessor measured every patient’s weight and reported it to his monthly blood tests, and make structured reports to the study
or her therapist.26 centre. Second, they were advised to admit patients to hospital
should they fall under a particular weight (body-mass index
Enhanced cognitive behaviour therapy
<14 kg/m²). Finally, they were informed about physical and
The unpublished German version of the manual for enhanced
psychiatric risks in patients with anorexia nervosa and were
cognitive behaviour therapy (CBT-E) used in this trial was
instructed to contact the respective study centre should a
developed in 2007 during initial training. It is based on a report
patient become at risk. In the study protocol, treatment
written by Fairburn before publication of his manual.25 Therapists
(dosage and type of therapy) in the optimised treatment as
in enhanced cognitive behaviour therapy have used Fairburn’s
usual study group was not regulated. Patients assigned to this
manual since its publication. The cognitive behaviour treatment
group had at least five contact sessions with the study centre,
plan consists of several modules, of which motivation (starting
at which their weight, laboratory findings, eating pathology,
well), nutrition, creating a formulation, and relapse prevention
and psychiatric comorbidity were investigated and monitored.
(ending well) are essential. Other modules target cognitive

www.thelancet.com Published online October 14, 2013 http://dx.doi.org/10.1016/S0140-6736(13)61746-8 3

 Articles

follow-up visit (appendix p 7). BMI measurements taken Statistical analysis

at the end of treatment and at the 12-month follow-up Our primary hypothesis had two parts. The first part
visit served as the ANTOP trial’s primary outcome. stated that the outpatient intervention of focal psycho­
Masked and trained assessors measured patients’ height dynamic therapy would have a better outcome with
and weight at the baseline assessment. Assessors respect to BMI at the end of treatment compared with
checked patients’ weight at every timepoint. Patients optimised treatment as usual. The second part stated that,
were weighed wearing light clothing (without shoes), compared with optimised treatment as usual, the
using a balance-beam scale that was recalibrated outpatient intervention of enhanced cognitive behaviour
regularly. We calculated BMI using bodyweight in kg therapy would have a better outcome with respect to BMI
divided by height in m squared (kg/m²). at the end of treatment.22 Before we began to obtain data,
Masked assessors measured secondary outcomes of we also proposed the same hypotheses for the 12-month
general psychopathology and eating disorder-specific follow-up visit. Additionally, we expected that both at the
psychopathology (appendix p 5). They used a self- end of treatment and at the 12-month follow-up visit,
assessment test—the eating disorder inventory-2 recovery rates defined in terms of a combined outcome
(EDI-2)—to investigate eating disorder pathology. We (in accordance with DSM-IV) would be higher for
used the structured clinical interview for DSM-IV axis I treatments specific for anorexia nervosa (ie, focal psycho­
mental disorders (SCID-I) to measure psychiatric dynamic therapy and enhanced cognitive behaviour
comorbidity. Furthermore, we used the full structured therapy) than for optimised treatment as usual.
interview for anorexic and bulimic syndromes (SIAB-EX) On the basis of the results of two smaller randomised
to measure in detail the symptomatology of anorexia controlled trials, in which the effects of focal psychodynamic
nervosa and bulimia nervosa. In addition to receiving therapy and enhanced cognitive behaviour therapy were
certified SCID-I training, the authors of the SIAB-EX assessed in outpatients with anorexia nervosa,13,14 we
interview held a 2-day workshop to train assessors how to assumed that the different interventions would result in
do expert ratings, before patient recruitment began. an improvement in BMI of 1·0 kg/m² compared with
Annual 1-day workshops were held throughout the study optimised treatment as usual (assumed SD 1·7; effect
period to ensure the high quality of data assessment. size 0·59). Because the hypothesis for the primary efficacy
Masked assessors also applied the psychiatric status endpoint entails two statistical tests—namely, focal
rating (PSR) scale based on the patient’s SIAB-EX psychodynamic therapy versus optimised treatment as
interview (appendix p 5). The PSR scale is used to usual, and enhanced cognitive behaviour therapy versus
measure the general severity of the anorexic disorder. optimised treatment as usual—the nominal α for the
PSR scores range from 1 (patient has no symptoms of primary hypothesis was set to 2·5% (two-sided) to restrict
anorexia nervosa) to 6 (patient has severe symptoms of the type 1 error to 5%. For a two-sided t test with 2·5%
anorexia nervosa that require admission). A score significance and 80% power, we needed a sample size of
of 5 indicates that all DSM-IV criteria for anorexia 55 patients per group, resulting in a study sample size
nervosa have been fulfilled. At baseline, all patients had a of 165. We increased the sample size to 242 patients to
PSR score of 4 (subsyndromal anorexia nervosa) or 5 (full allow for a dropout rate of at least 30%. The independent
syndromal anorexia nervosa). To ascertain a strictly coordination centre approved the statistical analysis plan
defined rating of outcome at the end of treatment and at before outcome data were examined.
the 12-month follow-up visit, we established a global We analysed the primary outcome by intention to treat,
outcome score based on the following combinations of which included all patients who underwent random­
PSR scores and BMI: recovery (scored as 3) was defined isation, at the end of treatment and at the 12-month
as a PSR score of 1 or 2 and BMI greater than 18·5 kg/m²; follow-up visit. We imputed missing data with a long­i­t­u­
full syndrome anorexia nervosa (scored as 1) was a PSR d­inal approach (mixed model for repeated measures
score of 5 or 6 and BMI of 17·5 kg/m² or lower; and [MMRM]). The ANTOP schedule of measure­ ments
partial syndrome anorexia nervosa (scored as 2) included provided a precondition for using MMRM—ie, one
all other cases. additional measurement timepoint between baseline and
We recorded service dosage in terms of the number of end of treatment—so time trends could be modelled. We
outpatient psychotherapy sessions accessed (includ­ decided to use MMRM as the first imputation method
ing the studied interventions of focal psychodynamic because findings of a series of studies showed that
therapy and enhanced cognitive behaviour therapy) and MMRM is more robust to biases from missing data than
use of any inpatient or day-patient treatment. Addition­ is the last-observation-carried-forward method.27 This
ally, we asked patients assigned to both focal psycho­ decision was noted in the statistical analysis plan before
dynamic therapy and enhanced cognitive behaviour data were examined. We also applied the mean-other
therapy groups to give brief feedback about the helpful­ imputation method. With this approach, missing values
ness of the sessions (gauged on a visual analogue scale are replaced with the mean of the other group to provide
from 0 to 10) and the length of treatment (too short, a conservative approach (in our trial, this process was
adequate, or too long). done to avoid erroneous decisions in favour of focal

4 www.thelancet.com Published online October 14, 2013 http://dx.doi.org/10.1016/S0140-6736(13)61746-8

 Articles

psychodynamic therapy or enhanced cognitive behaviour and SZ, WH, BW, GG, H-CF, and KEG were responsible
therapy; appendix p 8). As an additional sensitivity for submitting the manuscript. SZ made the final
measure, we did a complete case analysis (appendix p 9) decision to submit the paper for publication.
that included all patients who had BMI values at all
measurement timepoints. We also did a per-protocol Results
analysis that included all patients who received at least Between May, 2007, and June, 2009, we screened
27 therapy sessions, had a BMI measurement taken 727 patients for eligibility; 242 underwent randomisation
either at the end of treatment or at the 12-month after baseline assessment (figure 1). The number of
follow-up visit, were not pregnant during the trial or at patients enrolled per study centre was between 12 and 35.
12-month follow-up, and were not admitted to hospital Table 1 shows baseline characteristics. We did not record
for more than 4 weeks during the trial. any differences between groups with respect to demo­
For the primary outcome analysis, we compared both graphic characteristics, BMI, illness duration, subtype of
treatment groups (focal psychodynamic therapy and anorexia nervosa, and affective disorders. However, a
enhanced cognitive behaviour therapy) with the optimised comorbid anxiety disorder was more frequent in patients
treatment as usual group. The primary outcome—BMI allocated either enhanced cognitive behaviour therapy or
at the end of treatment—was analysed with a mixed optimised treatment as usual, compared with focal
modelling approach. We entered the grand mean, psycho­dynamic therapy (table 1). Overall, mean BMI at
treatment effects, and the binary stratification variable baseline was 16·7 kg/m² (SD 1·0) and mean weight was
(duration of anorexia nervosa, ≤6 years vs >6 years) as
fixed effects. Because trial sides were not chosen at
727 screened for eligibility*
random, we decided before data analysis started to model
the centre effect as a fixed effect. Baseline BMI was
entered as a covariate. The mathematical equation for our 485 excluded
197 did not meet eligibility criteria
modelling approach is described elsewhere.22 We tested 39 no current DSM-IV diagnosis of
the main hypothesis (primary outcome) by doing a series anorexia nervosa or subsyndromal
of pairwise comparisons. To investigate secondary anorexia nervosa
158 did not meet required weight range
hypotheses, we did exploratory analyses with a similar 157 did not meet exclusion criteria
approach. In all analyses, we entered the variable of 45 psychiatric exclusion
16 medical exclusion
centre as a control variable. We analysed the global 37 lived too far away
outcome with the MMRM approach, and this variable 52 ongoing psychotherapy
was treated as continuous. For the moderator analysis of 2 pregnancy
5 other reasons
anorexia nervosa subtypes, we divided the study sample 127 patients declined participation
into two groups and did the MMRM analysis for each 4 unrecorded
group separately. We set the significance level to 5% for
exploratory analyses. We used SAS versions 9.1 and 9.2 242 randomly allocated to treatment groups
for statistical analyses.
In patients with anorexia nervosa, food restriction and
purging behaviour can lead to life-threatening starvation
that requires inpatient medical monitoring. Because 80 assigned to receive focal 80 assigned to receive enhanced 82 assigned to receive optimised
psychodynamic therapy cognitive behaviour therapy treatment as usual
severe weight loss is central to the psychopathology of 53 completed treatment 65 completed treatment
anorexia nervosa, intermittent inpatient treatment of up
to 4 weeks as a crisis intervention was not judged a serious
adverse event. All other life-threatening or fatal events 3 months after treatment start 3 months after treatment start 3 months after treatment start
71 assessed 74 assessed 66 assessed
were defined as serious adverse events, and these had to 9 lost to follow-up 6 lost to follow-up 16 lost to follow-up
be reported immediately to the principal investigator. End of treatment End of treatment End of treatment
Additionally, we set up an independent safety and data 63 assessed 72 assessed 53 assessed
monitoring board. This board consisted of internationally 17 lost to follow-up 8 lost to follow-up 29 lost to follow-up

renowned experts in the area of eating disorder research 3-month follow-up 3-month follow-up 3-month follow-up
57 assessed 66 assessed 48 assessed
and treatment, and in data and safety monitoring. Further 23 lost to follow-up 14 lost to follow-up 34 lost to follow-up
details about medical complications in the ANTOP trial 12-month follow-up 12-month follow-up 12-month follow-up
have been published elsewhere.22 This trial is registered at 58 assessed 65 assessed 46 assessed
http://isrctn.org, number ISRCTN72809357. 22 lost to follow-up 15 lost to follow-up 36 lost to follow-up

Role of the funding source 80 analysed for primary outcome 80 analysed for primary outcome 82 analysed for primary outcome
The sponsors of the study had no role in study design,
data collection, data analysis, data interpretation, or Figure 1: Trial profile
writing of the report. All authors had access to the data, *Six male patients were excluded before screening because of predefined inclusion criteria.

www.thelancet.com Published online October 14, 2013 http://dx.doi.org/10.1016/S0140-6736(13)61746-8 5

 Articles

46·5 kg (SD 4·2). 94 (39%) patients had anorexia nervosa end of treatment, patients in all study groups showed
for longer than 6 years. A restrictive subtype of anorexia substantial weight gains from baseline (focal psycho­
nervosa was present in 131 (53%) patients, and 96 (40%) dynamic therapy, 0·73 kg/m²; enhanced cognitive behav­
had at least one additional SCID-I diagnosis of a co­ iour therapy, 0·93 kg/m²; optimised treatment as usual,
morbid mental disorder. 0·69 kg/m²). We recorded no differences in BMI between
After 10 months of treatment (at the end of treatment), study groups at the end of treatment in the adjusted
54 (22%) of 242 patients were lost to follow-up, and models (table 2), using the MMRM algorithm to replace
73 (30%) had dropped out by 12-month follow-up. At both missing values. At 12-month follow-up, mean BMI values
these timepoints, rates of loss-to-follow-up differed for patients from all study groups had risen further (focal
significantly between study groups, with the highest rate psychodynamic therapy, 1·64 kg/m²; en­hanced cognitive
noted in the optimised treatment as usual group. We did behaviour therapy, 1·30 kg/m²; optim­ised treatment as
a sensitivity analysis to detect any possible selection bias usual, 1·22 kg/m²). Again, we did not note a significant
attributable to different dropout rates at the end of difference in BMI between study groups (table 2). A
treatment and at 12-month follow-up; BMI at baseline, sensitivity analysis using the mean-other impu­ tation
difference in BMI between baseline and the end of treat­ method showed the same result pattern as with the
ment, anorexia nervosa subtype, and comorbid diagnosis MMRM approach (appendix, p 8). Complete case analysis
of a mental disorder were not related to the dropout rate. (appendix, p 9) and per-protocol analysis (data not shown)
Table 2 shows outcome data after 4 months of treat­ did not show any different results.
ment, at the end of treatment, at 3-month follow-up, and Exploratory data analyses were done to investigate the
at 12-month follow-up. Figure 2 depicts the course of proportion of patients with full and partial anorexia
weight gain during treatment and at follow-up. At the nervosa syndrome at baseline and those showing full
recovery at the end of treatment and at 12-month follow-up
(figure 3). Study groups did not differ in terms of global
Focal psycho­ Enhanced cognitive Optimised
dynamic therapy behaviour therapy treatment
outcome between baseline and the end of treat­ment. At
(n=80) (n=80) as usual 12-month follow-up, however, patients assigned focal
(n=82) psychodynamic therapy had a significantly higher recovery
Demographic characteristics rate compared with optimised treatment as usual (full
Mean (SD) age at entry (years) 28·0 (8·6) 27·4 (7·9) 27·7 (8·1) recovery, 35% vs 13%; p=0·036). Table 3 shows BMI-related
Marital status within-group effect sizes and table 4 provides additional
Single, never married 65 (81%) 66 (83%) 66 (81%) information for the 12-month follow-up outcome.
Married, or living as such 12 (15%) 7 (9%) 13 (16%) Because fewer patients allocated focal psychodynamic
Separated or divorced 3 (4%) 5 (6%) 3 (4%) therapy had comorbid anxiety disorders at baseline,
Unknown 0 2 (3%) 0 compared with the other treatment groups, we did a
Clinical characteristics sensitivity analysis to investigate whether an anxiety
Mean (SD) weight (kg) 46·37 (4·3) 46·33 (3·9) 46·71 (4·4) disorder at baseline could have affected the BMI outcome
Mean (SD) body-mass index (kg/m²) 16·57 (1·0) 16·82 (1·0) 16·75 (1·0)
at the end of treatment or at 12-month follow-up. Results
Body-mass index
of the MMRM analysis showed no such association.
<17·5 kg/m² 62 (78%) 53 (66%) 56 (68%)
Further subgroup analyses were done of patients with
17·5 to ≤18·5 kg/m² 18 (23%) 27 (34%) 26 (32%)
baseline BMI less than 17·5 kg/m², which accords with the
weight criterion for full syndrome anorexia nervosa. In
Duration of illness
this subgroup, at the end of treatment, mean BMI in
≤6 years 49 (61%) 49 (61%) 50 (61%)
patients assigned focal psychodynamic therapy was
>6 years 31 (39%) 31 (39%) 32 (39%)
lower than in those allocated enhanced cognitive behav­
Anorexia nervosa subtypes
iour therapy (16·9 kg/m² vs 17·5 kg/m²; p=0·038). Analysis
Binge-purge 34 (43%) 38 (48%) 39 (48%)
of anorexia nervosa subtypes (restrictive vs binge-purge)
Restrictive 46 (58%) 42 (53%) 43 (52%)
showed no differences in treatment response between
Comorbidities
these two intervention groups.
Affective disorder 14 (18%) 25 (31%) 19 (23%)
Eating disorder psychopathology with respect to self-
Anxiety disorder 11 (14%) 20 (25%) 28 (34%)
rating (EDI-2) did not differ over the course of treatment
Somatoform disorder 1 (1%) 3 (4%) 1 (1%)
and follow-up (table 2). However, with the expert inter­
Substance abuse 0 0 0 view (SIAB-EX), patients with anorexia nervosa who were
Mean (SD) total score on the eating disorder inventory 256 (54·6) 271 (53·4) 275 (51·7) assigned enhanced cognitive behaviour therapy had the
Mean (SD) total score on the structured inventory for 1·0 (0·3) 1·1 (0·3) 1·1 (0·3) lowest SIAB-EX scores at the end of treatment (table 2).
anorexic and bulimic syndromes
At 12-month follow-up, however, this difference was no
Data are mean (SD) or number of patients (%). longer detectable.
No serious adverse events attributable to weight loss
Table 1: Baseline characteristics
or trial participation were reported during the study.

6 www.thelancet.com Published online October 14, 2013 http://dx.doi.org/10.1016/S0140-6736(13)61746-8

 Articles

Between baseline and 12-month follow-up, 13 (23%) of (30·6) for optimised treatment as usual (p=0·49). For
57 patients assigned focal psychodynamic therapy (with admissions between baseline and 12-month follow-up,
available data), 20 (34%) of 58 allocated enhanced cog­ mean duration was 19·0 days (SD 52·7) for focal
nitive behaviour therapy, and 21 (41%) of 51 assigned psychodynamic therapy, 29·4 days (55·3) for enhanced
optimised treatment as usual received additional in­ cognitive behaviour therapy, and 29·3 days (54·2) for
patient treatment. The proportion receiving treatment optimised treatment as usual (p=0·52). However, the
differed significantly between the focal psychodynamic distributions of days in hospital were skewed such that a
therapy group and the optimised treatment as usual few patients had a comparably long duration and more
group (p=0·044), whereas other group comparisons did patients had short durations.
not differ by much. Between baseline and the end of The overall dosage of outpatient psychotherapy ses­
treatment, two patients assigned focal psychodynamic sions did not differ from baseline to 12-month follow-up
therapy, three allocated enhanced cognitive behaviour between treatment groups (focal psychodynamic therapy,
therapy, and five assigned optimised treatment as usual mean 39·9 sessions, 95% CI 33·8–46·5; enhanced cog­
had inpatient treatment due to weight loss for 28 days or nitive behaviour therapy, 44·8, 38·4–50·8; optimised
less; inpatient treatment for longer than 28 days was treatment as usual, 41·6, 35·1–48·1; p=0·503). At the end
given to five patients assigned focal psychodynamic of treatment, 110 (81%) of 135 participants who were
therapy, eight allocated enhanced cognitive behaviour assessed from the focal psychodynamic therapy and
therapy, and nine assigned optimised treatment as usual. enhanced cognitive behaviour therapy groups gave full or
The mean duration of admissions that arose between partial feedback about their treatment. On a scale of 0
baseline and the end of treatment was 6·8 days (SD 22·9) (therapy was not at all helpful) to 10 (therapy was very
for focal psychodynamic therapy, 12·6 days (36·9) for helpful), mean values were reported of 7·3 (SD 2·6) for
enhanced cognitive behaviour therapy, and 12·5 days focal psychodynamic therapy and 7·6 (2·3) for enhanced

Focal psycho­ Enhanced cognitive Optimised Focal psychodynamic therapy Focal psychodynamic therapy vs Enhanced cognitive behaviour
dynamic therapy behaviour therapy treatment as usual vs enhanced cognitive optimised treatment as usual therapy vs optimised treatment
behaviour therapy as usual
Ls-mean 95% CI Ls-mean 95% CI Ls-mean 95% CI Ls-m diff p Effect Ls-m diff p Effect Ls-m p Effect
(SE) (SE) (SE) (95% CI) size (95% CI) size diff (95% CI) size
Body-mass index
After 4 months 16·94 16·67– 17·08 16·81– 16·96 16·68– −0·14 (−0·51 0·46 −0·12 −0·01 0·94 −0·01 0·12 0·52 0·11
of treatment (0·14) 17·21 (0·13) 17·34 (0·14) 17·23 to 0·23) (−0·39 to 0·36) (−0·25 to 0·50)
After 10 months 17·30 16·92– 17·75 17·39– 17·44 17·05– –0·45 0·09 −0·29 −0·14 0·60 −0·09 0·3 0·26 0·20
of treatment (end (0·19) 17·67 (0·18) 18·11 (0·20) 17·83 (−0·96 to 0·07) (−0·68 to 0·39) (−0·22 to 0·83)
of treatment)
At 3-month 17·63 17·20– 17·74 17·33– 17·74 17·29– −0·11 0·70 −0·07 −0·11 0·72 −0·07 0 1·00 0·00
follow-up (0·22) 18·05 (0·21) 18·15 (0·23) 18·19 (−0·70 to 0·47) (−0·73 to 0·51) (−0·60 to 0·60)
At 12-month 18·20 17·72– 18·10 17·64– 17·95 17·44– 0·10 0·76 0·05 0·25 0·48 0·13 0·15 0·67 0·08
follow-up (0·24) 18·69 (0·23) 18·56 (0·26) 18·47 (−0·56 to 0·76) (−0·45 to 0·95) (−0·54 to 0·83)
EDI, total score
After 4 months 295 286– 295 287– 293 284– −0·27 0·97 −0·01 1·81 0·78 0·05 2·08 0·74 0·06
of treatment (4·46) 304 (4·32) 304 (4·57) 302 (−12·30 to 11·77) (−10·68 to 14·30) (−10·08 to 14·24)
After 10 months 272 260– 270 259– 277 264– 1·67 0·84 0·03 −4·98 0·58 −0·10 −6·64 0·44 −0·14
of treatment (end (6·18) 284 (5·83) 282 (6·46) 289 (−14·92 to 18·26) (−22·53 to 12·58) (−23·63 to 10·34)
of treatment)
At 3-month 269 257– 270 258– 274 260– −0·89 0·92 −0·02 −4·41 0·64 −0·09 −3·52 0·70 −0·07
follow-up (6·41) 282 (6·03) 282 (6·76) 287 (−18·10 to 16·32) (−22·71 to 13·90) (−21·23 to 14·19)
At 12-month 257 244– 263 251– 260 246– −6·17 0·51 −0·12 −2·98 0·76 −0·06 3·19 0·74 0·06
follow-up (6·76) 271 (6·47) 276 (7·22) 275 (−24·49 to 12·15) (−22·42 to 16·47) (−15·79 to 22·17)
SIAB-EX, total score
After 10 months 0·86 0·77– 0·77 0·68– 0·89 0·80– 0·09 0·14 0·26 −0·03 0·61 −0·09 −0·12 0·05 −0·35
of treatment (end (0·05) 0·95 (0·04) 0·85 (0·05) 0·98 (−0·03 to 0·21) (−0·16 to 0·09) (−0·25 to −0·00)
of treatment)
At 12-month 0·72 0·61– 0·73 0·63– 0·71 0·59– −0·01 0·88 −0·03 0·01 0·92 0·02 0·02 0·81 0·05
follow-up (0·05) 0·82 (0·05) 0·83 (0·06) 0·82 (−0·15 to 0·13) (−0·15 to 0·16) (−0·13 to 0·17)

Differences between groups were tested using the final adjusted models; missing values were replaced by the mixed model for repeated measures algorithm. 95% CIs are given for estimated least square means
(Ls-mean) for every treatment group and for least square mean differences (Ls-m diff) between treatment groups. EDI=eating disorder inventory. SIAB-EX=structured inventory of anorexic and bulimic
syndromes, expert version.

Table 2: Adjusted mean scores for body-mass index and eating disorder pathology, by treatment group

www.thelancet.com Published online October 14, 2013 http://dx.doi.org/10.1016/S0140-6736(13)61746-8 7

 Articles

18·5 Focal psychodynamic therapy

cognitive behaviour therapy. 44% of patients assigned
Enhanced cognitive behaviour therapy focal psychodynamic therapy found the length of treat­
18·3 Optimised treatment as usual ment adequate, 52% said it was too short, and 4% judged
End of treatment it too long; by comparison, 40% of patients allocated
18·1
enhanced cognitive behaviour therapy said the treatment
17·9
length was adequate, 49% thought it was too short, and
Body-mass index (kg/m2)

17·7 11% judged it too long.

17·5
End of follow-up Discussion
17·3 Findings of the ANTOP study show that outpatient
17·1
treatment of adults with anorexia nervosa by either
optimised treatment as usual, focal psychodynamic
16·9 therapy, or enhanced cognitive-behaviour therapy leads
16·7 to relevant weight gains and a decrease in general and
eating disorder-specific psychopathology during the
16·5
Baseline 4 months 10 months 3-month 12-month course of treatment. These positive effects continue
of treatment of treatment follow-up follow-up beyond treatment until 12-month follow-up. However,
Measurement timepoints
the primary hypothesis of the ANTOP study was not
Figure 2: Course of weight gain during treatment and follow-up, by treatment group confirmed: no difference in weight gain was recorded by
Data are least square means (Ls-mean). Error bars show SE. the end of treatment between the study groups. Weight
gains noted in the ANTOP study accord with those
reported in small single-centre studies.16,17,20 However,
with respect to the global outcome measure, patients
Recovered anorexia nervosa Partial syndrome anorexia nervosa Full syndrome anorexia nervosa allocated focal psychodynamic therapy had higher
100 recovery rates compared with those assigned optimised
90
treatment as usual at 12-month follow-up.
80
Under close guidance from their family doctor—eg,
Proportion of patients (%)

70
regular weight monitoring and essential blood testing—
60
and with close supervision of their respective study
50
centre, patients allocated optimised treatment as usual
40
were able to choose their favourite treatment approach
30
and setting (intensity, inpatient, day patient, or out­
20
patient treatment) and their therapist, in accordance
10
with German national treatment guidelines for anorexia
0
Baseline End of 12-month Baseline End of 12-month Baseline End of 12-month nervosa.11 Moreover, comparisons of applied dosage and
treatment follow-up treatment follow-up treatment follow-up
intensity of treatment showed that all patients—
Focal psychodynamic therapy Enhanced cognitive Optimised treatment as usual irrespective of treatment allocation—averaged a similar
behaviour therapy number of outpatient sessions over the course of the
Figure 3: Recovery rates during treatment and follow-up, by treatment group
treatment and follow-up periods (about 40 sessions).
These data partly reflect an important achievement of
the German health-care system: that access to psycho­
therapy treatment is covered by insurance. However,
Focal psychodynamic Enhanced cognitive Optimised treatment
patients allocated optimised treatment as usual needed
therapy behaviour therapy as usual additional inpatient treatment more frequently (41%)
Effect size p Effect size p Effect size p
than either those assigned focal psychodynamic therapy
(95% CI) (95% CI) (95% CI) (23%) or enhanced cognitive behaviour therapy (35%).
After 4 months of treatment 0·23 0·12 0·37 0·003 0·23 0·10
In a study in adolescent patients, similar results were
(0·06–0·50) (0·12–0·61) (0·04–0·52) noted over a 2-year period, with fewer admissions
After 10 months of 0·62 0·0003 1·00 <0·0001 0·71 0·0003 reported with family-based treat­ment (15%) compared
treatment (end of treatment) (0·29–0·90) (0·60–1·41) (0·35–1·10) with a more general approach of adolescent-focused
At 3-month follow-up 0·95 <0·0001 1·00 <0·0001 1·00 <0·0001 therapy (37%).28 Although, to date, clear consensus has
(0·56–1·28) (0·54–1·48) (0·60–1·46) not been reached with respect to the definition of
At 12-month follow-up 1·60 <0·0001 1·40 <0·0001 1·20 <0·0001 outcome in anorexia nervosa,29 analysis of recovery rates
(1·10–2·00) (0·87–1·87) (0·74–1·77)
is common in psychotherapy trials.30 Our results showed
Data are effect size (95% CI), standardised by the mixed model for repeated measures. that at follow up, patients assigned focal psychodynamic
therapy had significantly higher recovery rates compared
Table 3: Within-group changes in body-mass index from baseline, by treatment group
with those receiving optimised treatment as usual.

8 www.thelancet.com Published online October 14, 2013 http://dx.doi.org/10.1016/S0140-6736(13)61746-8

 Articles

Psychodynamic treatments make interpersonal

Focal psychodynamic Cognitive behaviour Treatment as F or χ² p
relationships the major theme. Compared with cognitive therapy (n=58) therapy (n=65) usual (n=46)
behaviour therapy they are less directive, induce
Mean (SD) weight (kg) 51·3 (7·2) 51·0 (8·8) 48·6 (9·7) 1·60 0·20
augmented emotional arousal, and target insight more
Body-mass index
(vs behaviour and cognition).31 In view of difficulties in the
≤17·5 kg/m² 18 (31%) 22 (34%) 19 (41%) 1·24 0·54
field of autonomy that individuals with anorexia nervosa
>17 5 kg/m² 40 (69%) 43 (66%) 27 (59%)
have, we postulate that these specific aspects of psycho­
Full criteria for anorexia 12 (21%) 14 (22%) 13 (28%) 0·97 0·62
dynamic therapy contribute to the positive effects of treat­ nervosa*
ment in this patient group. Substantial and lasting Comorbidities
changes after interpersonal treatment for eating disorders
Affective disorder 7 (12%) 10 (15%) 4 (9%) 1·12 0·57
have been shown for bulimia nervosa.32 Findings of a New
Anxiety disorder 5 (9%) 8 (12%) 9 (20%) 2·76 0·25
Zealand study17,18 indicate that psychotherapy focusing on
Somatoform disorder 1 (2%) 2 (3%) 0 1·46 0·48
interpersonal aspects (eg, interpersonal therapy and focal
Substance abuse 0 1 (2%) 0 1·61 0·45
psychodynamic therapy) across different eating dis­
Bulimia nervosa 4 (7%) 4 (6%) 3 (7%) 0·03 0·99
orders might need a prolonged timeframe to show effects,
thereby producing better long-term results.18,32 In a single- Data are number of patients (%), unless otherwise stated. F values (and corresponding p values) are derived from
centre study comparing two manual-based treatments in ANOVA tests to compare the three groups. χ2 values (and corresponding p values) compare percentages between the
three groups. *Patients with a body-mass index of ≤17·5 kg/m² and psychiatric status rating score of 5 or 6 met full
anorexia nervosa outpatients,30 significant weight gains criteria for anorexia nervosa.
were noted for patients in both study groups, with no
differences recorded between the two approaches with Table 4: Clinical characteristics at 12-month follow-up
respect to the amount of weight gained. However, these
researchers showed that the estimated proportion of
recovered anorexia nervosa patients depended strongly on
the underlying definition of recovery. When applying a Panel 2: Research in context
strict definition of recovery, as we did in our study
(a combination of objective measure [weight] and eating Systematic review
disorder pathology based on expert assessment), recovery We searched PubMed and the Cochrane Library for full papers
rates were 7–13% at 6 months and 14–19% at 12 months. published before May 3, 2013, reporting randomised controlled
Again, no differences with respect to recovery rates were trials, systematic reviews, and meta-analyses, with the MeSH
noted between treatments. terms: “anorexia nervosa”, “treatment”, “psychotherapy”,
Treatment acceptance is a major challenge in the “cognitive behavior therapy”, and “psychodynamic therapy”.
management of patients with anorexia nervosa.33 The We did not apply any language restrictions. We excluded trials
main reason for this difficulty is typically the patient’s focused exclusively on adolescents, group interventions, and
high ambivalence and lack of acceptance of the serious­ family therapy and those reporting on pure education. Our
ness of their illness. Thus, therapists not only have to search identified five systematic reviews,1,3,29 one meta-analysis,13
cope with a frequently difficult therapy process but also and three additional trials that were not included in the
must take responsibility for management of physical and respective reviews.16,20,30 To date, no evidence provides solid
psychiatric complications of this potentially lethal support for a particular therapeutic approach, setting, or
disorder. In the ANTOP study, we had a clear framework procedure for treatment of adults with anorexia nervosa. Thus,
of rules for medical and psychiatric monitoring and a large, multicentre, randomised controlled trials of commonly
minimum weight limit for admission to short-term in­ used psychotherapies for older adolescents and adults with
patient treatment (BMI <14 kg/m²). In both manual- anorexia nervosa are needed urgently.
based treatments, we included a brief nutrition guideline, Interpretation
and a structured session for close family and other The findings of the ANTOP trial showed that multicentre
relatives was also offered. These design aspects of the outpatient studies in patients with anorexia nervosa are
ANTOP study contributed to relatively high treatment possible. We also showed that patients can be treated
completion rates (76% average; 70% with focal psycho­ safely, many individuals with anorexia nervosa gain weight,
dynamic therapy; 81% with enhanced cognitive behaviour and that a substantial proportion have striking
therapy) compared with other small-scale adult anorexia improvements in eating disorder pathology and comorbid
nervosa studies.10,15 Moreover, patients allocated either psychopathology. The findings of the ANTOP study provide
focal psychodynamic therapy or enhanced cognitive evidence to support use of manual-based interventions;
behaviour therapy gave positive ratings to their treatment focal psychodynamic therapy proved most advantageous in
experience, which might have contributed further to the terms of recovery at 12-month follow-up, and enhanced
comparably low dropout rates. cognitive behaviour therapy was most effective in terms of
Previous findings show that a comorbid anxiety the speed of weight gain and improvements in eating
disorder could be associated with a poorer treatment disorder psychopathology.
outcome in individuals with anorexia nervosa.34,35 In the

www.thelancet.com Published online October 14, 2013 http://dx.doi.org/10.1016/S0140-6736(13)61746-8 9

 Articles

ANTOP study sample, despite the randomisation algo­ Treatment leaders for focal psychodynamic therapy were HS, H-CF, and
rithm, fewer patients allocated focal psychodynamic WH, for enhanced cognitive behaviour therapy were GG, MdZ, and MT
(with initial support from C Fairburn), and for optimised treatment as
therapy had comorbid anxiety disorders at baseline com­ usual were SZ, GG, MT, KEG, H-CF, and WH. Treatment leaders
pared with those assigned enhanced cognitive behaviour designed the treatment manuals in collaboration with the principal
therapy and optimised treatment as usual. However, investigators, and trained and supervised the trial therapists. The writing
results of a sensitivity analysis showed no association and publication oversight committee wrote the report. All authors
commented on drafts and approved the final version.
between anxiety disorders at baseline and BMI at the end
of treatment or at 12-month follow-up. ANTOP study group
Trial management group—SZ (chair), BW, GG, H-CF, MT, DS, KEG,
The ANTOP study was designed as a large, multicentre, MdZ, AD, SH, MB, BL, ST, JvW, AZ, CS-B, HS, and WH (co-chair); trial
randomised controlled trial in adults with anorexia steering committee—SZ (chair) and WH (co-chair); data monitoring and
nervosa. These design features overcome the short­com­ ethics committee—U Schmidt (London, UK), H-C Deter (Berlin,
ings of previous studies (panel 2) by providing a random­ Germany), S Schneider (Bochum, Germany), and A Faldum (Münster,
Germany); analysis strategy group—BW (chair), DS, WH, KEG, GG, SZ,
ised controlled design, a large sample size, appro­priate and CS-B; writing and publication oversight committee—SZ (co-chair), BW
inclusion criteria, a detailed treatment proto­col, and a (co-chair), GG, H-CF, MT, DS, MdZ, and KEG; trial manager—GG.
clear separation of intervention conditions.22 We judged Conflicts of interest
our study a success because most patients completed We declare that we have no conflicts of interest.
treatment and reported high satisfaction scores with the Acknowledgments
treatments; the dropout rate was much lower than in The German Federal Ministry of Education and Research
previous anorexia nervosa treatment studies.10,17 (Bundesministerium für Bildung und Forschung [BMBF], project
number 01GV0624) funded the ANTOP study, which is part of the
Our trial, however, had several limitations. First, we
BMBF research programme Research Networks on Psychotherapy. The
chose gain in BMI as a simple, objective, and conservative ANTOP study was designed at the Department of Psychosomatic
primary outcome measure. This approach might have Medicine and Psychotherapy, University Hospital Tübingen, and the
been too one-dimensional. To get a more comprehensive Department of General Internal Medicine and Psychosomatics,
Heidelberg University Hospital. The University of Tübingen undertook
picture, we also looked at a combined secondary outcome
the responsibilities of sponsor in terms of the guidelines of good
that included core aspects of eating disorder psycho­ clinical practice in clinical trials (ICH-GCP, E6); the sponsor declaration
pathology. Second, although the dropout rate within our was signed by the Dean of the Medical Faculty. Data management was
overall sample was acceptable, we noted a higher provided by the Coordination Center for Clinical Trials, Marburg (CS-B).
F Schmid (University Hospital Erlangen) was responsible for
dropout rate for patients assigned optimised treatment
independent data monitoring. C Fairburn (Oxford, UK) provided the
as usual compared with the other groups. Therefore, initial 2 days of training for enhanced cognitive behaviour therapy and
infor­mation on the course of the illness in individuals in the provisional manual, before its publication in English and German,
this group is limited. We only had scant contact with but was not involved in any further conduct of the ANTOP study. We
thank Doro Niehoff for help with data management and
patients assigned optimised treatment as usual at our Nichole Martinson for editorial assistance; the participants who took
centres and, thus, formed a weaker alliance with part in the ANTOP study; and the therapists and staff from all study
these patients. This dimin­ished personal contact might centres, who helped with patient recruitment, diagnostic procedures,
have contributed to the dropout rate. Finally, we were and monitoring.
restricted by funding to intermediate follow-up of References
12 months. Additional long-term follow-up measurement 1 Treasure J, Claudino AM, Zucker N. Eating disorders. Lancet 2010;
375: 583–93.
points for outcome data (for at least 5 years after initial 2 Herzog W, Deter HC, Fiehn W, Petzold E. Medical findings and
treatment ends) are necessary. predictors of long-term physical outcome in anorexia nervosa:
Although the findings of the ANTOP study suggest a prospective, 12-year follow-up study. Psychol Med 1997; 27: 269–79.
3 Fairburn CG, Harrison PJ. Eating disorders. Lancet 2003; 361: 407–16.
that adults with anorexia nervosa have a realistic chance
4 Arcelus J, Mitchell AJ, Wales J, Nielsen S. Mortality rates in patients
of recovery or, at least, can achieve substantial improve­ with anorexia nervosa and other eating disorders: a meta-analysis of
ment, a relevant proportion of patients still had anorexic 36 studies. Arch Gen Psychiatry 2011; 68: 724–31.
symptoms at the end of our study. Anorexia nervosa 5 Zipfel S, Löwe B, Reas DL, Deter H-C, Herzog W. Long-term
prognosis in anorexia nervosa: lessons from a 21-year follow-up
“remains an enigma and its clinical challenge is [still] study. Lancet 2000; 355: 721–22.
intimidating”.36 Besides strategies of prevention and early 6 Klump KL, Bulik CM, Kaye WH, Treasure J, Tyson E. Academy for
intervention, we have to refine our treatments further to eating disorders position paper: eating disorders are serious mental
illnesses. Int J Eat Disord 2009; 42: 97–103.
fight the vicious cycles of dieting behaviour.37 7 Herzog W, Schellberg D, Deter HC. First recovery in anorexia
Contributors nervosa patients in the long-term course: a discrete-time survival
The principal investigators (SZ, WH, BW, and GG) designed the study analysis. J Consult Clin Psychol 1997; 65: 169–77.
and obtained funding. The ANTOP trial management group, trial 8 Stuhldreher N, Konnopka A, Wild B, et al. Cost-of-illness studies
steering committee, and the data monitoring and ethics committee and cost-effectiveness analyses in eating disorders: a systematic
further developed study design. The statistical analysis plan was written review. Int J Eat Disord 2012; 45: 476–91.
by the analysis strategy group and approved by the trial steering 9 Smink FR, van Hoeken D, Hoek HW. Epidemiology of eating
committee and the data monitoring and ethics committee before the disorders: incidence, prevalence and mortality rates.
analysis began. DS and BW did the main statistical analysis. Patient Curr Psychiatry Rep 2012; 14: 406–14.
recruitment was done by SH (Bochum), MdZ (Hannover), Prof W Senf 10 Halmi KA, Agras WS, Crow S, et al. Predictors of treatment
(Essen), AZ (Freiburg), BL (Hamburg), WH (Heidelberg), acceptance and completion in anorexia nervosa: implications for
future study designs. Arch Gen Psychiatry 2005; 62: 776–81.
Prof P Henningsen (Munich), SZ (Tübingen), and JvW (Ulm).

10 www.thelancet.com Published online October 14, 2013 http://dx.doi.org/10.1016/S0140-6736(13)61746-8

 Articles

11 Herpertz S, Hagenah U, Vocks S, von Wietersheim J, Cuntz U, 24 Nordle O, Brantmark B. A self-adjusting randomization plan for
Zeeck A, and the German Society of Psychosomatic Medicine and allocation of patients into two treatment groups.
Psychotherapy, and the German College for Psychosomatic Clin Pharmacol Ther 1977; 22: 825–30.
Medicine. The diagnosis and treatment of eating disorders. 25 Fairburn CG. Cognitive behavior therapy and eating disorders.
Dtsch Arztebl Int 2011; 108: 678–85. New York: The Guilford Press, 2008.
12 NICE. Eating disorders: core interventions in the treatment and 26 Schauenburg H, Friederich H-C, Wild B, Zipfel S, Herzog W. Focal
management of anorexia nervosa, bulimia nervosa and related psychodynamic psychotherapy of anorexia nervosa: a treatment
eating disorders. Jan 28, 2004. http://www.nice.org.uk/nicemedia/ manual. Psychotherapeut 2009; 54: 270–80 (in German).
live/10932/29220/29220.pdf (accessed Sept 5, 2013). 27 Mallinckrodt CH, Watkin JG, Molenberghs G, Carroll RJ. Choice of
13 Hay PP, Bacaltchuk J, Byrnes RT, Claudino AM, Ekmejian AA, the primary analysis in longitudinal clinical trials. Pharm Stat 2004;
Yong PY. Individual psychotherapy in the outpatient treatment of 3: 161–69.
adults with anorexia nervosa. Cochrane Database Syst Rev 2009; 28 Lock J, Le Grange D, Agras WS, Moye A, Bryson SW, Jo B.
published online Jan 21. DOI:10.1002/14651858.CD003909. Randomized clinical trial comparing family-based treatment with
14 Dare C, Eisler I, Russell G, Treasure J, Dodge L. Psychological adolescent-focused individual therapy for adolescents with anorexia
therapies for adults with anorexia nervosa: randomised controlled nervosa. Arch Gen Psychiatry 2010; 67: 1025–32.
trial of out-patient treatments. Br J Psychiatry 2001; 178: 216–21. 29 Watson HJ, Bulik CM. Update on the treatment of anorexia
15 Pike KM, Walsh BT, Vitousek K, Wilson GT, Bauer J. Cognitive nervosa: review of clinical trials, practice guidelines and emerging
behavior therapy in the posthospitalization treatment of anorexia interventions. Psychol Med 2012; 10: 1–24.
nervosa. Am J Psychiatry 2003; 160: 2046–49. 30 Schmidt U, Oldershaw A, Jichi F, et al. Out-patient psychological
16 Fairburn CG, Cooper Z, Doll HA, O’Connor ME, Palmer RL, therapies for adults with anorexia nervosa: randomised controlled
Dalle Grave R. Enhanced cognitive behaviour therapy for adults trial. Br J Psychiatry 2012; 201: 392–99.
with anorexia nervosa: a UK-Italy study. Behav Res Ther 2013; 31 Malik ML, Beutler LE, Alimohamed S, Gallagher-Thompson D,
51: R2–8. Thompson L. Are all cognitive therapies alike? A comparison of
17 McIntosh VV, Jordan J, Carter FA, et al. Three psychotherapies for cognitive and noncognitive therapy process and implications for the
anorexia nervosa: a randomized, controlled trial. Am J Psychiatry application of empirically supported treatments.
2005; 162: 741–47. J Consult Clin Psychol 2003; 71: 150–58.
18 Carter FA, Jordan J, McIntosh VV, et al. The long-term efficacy of 32 Fairburn CG, Norman PA, Welch SL, O’Connor ME, Doll HA,
three psychotherapies for anorexia nervosa: a randomized, Peveler RC. A prospective study of outcome in bulimia nervosa and
controlled trial. Int J Eat Disord 2011; 44: 647–54. the long-term effects of three psychological treatments.
19 Lipsman N, Woodside DB, Giacobbe P, et al. Subcallosal cingulate Arch Gen Psychiatry 1995; 52: 304–12.
deep brain stimulation for treatment-refractory anorexia nervosa: 33 Dejong H, Broadbent H, Schmidt U. A systematic review of
a phase 1 pilot trial. Lancet 2013; 381: 1361–70. dropout from treatment in outpatients with anorexia nervosa.
20 Touyz S, Le Grange D, Lacey H, et al. Treating severe and Int J Eat Disord 2012; 45: 635–47.
enduring anorexia nervosa: a randomized controlled trial. 34 Zerwas S, Lund BC, Von Holle A, et al. Factors associated with
Psychol Med 2013; published online May 3. DOI:10.1017/ recovery from anorexia nervosa. J Psychiatr Res 2013; 47: 972–79.
S0033291713000949. 35 Kaye WH, Bulik CM, Thornton L, Barbarich N, Masters K.
21 Flament MF, Bissada H, Spettigue W. Evidence-based Comorbidity of anxiety disorders with anorexia and bulimia
pharmacotherapy of eating disorders. Int J Neuropsychopharmacol nervosa. Am J Psychiatry 2004; 161: 2215–21.
2012; 15: 189–207. 36 Strober M, Johnson C. The need for complex ideas in anorexia
22 Wild B, Friederich HC, Gross G, et al. The ANTOP study: nervosa: why biology, environment, and psyche all matter, why
focal psychodynamic psychotherapy, cognitive-behavioural therapy, therapists make mistakes, and why clinical benchmarks are needed
and treatment-as-usual in outpatients with anorexia for managing weight correction. Int J Eat Disord 2012; 45: 155–78.
nervosa—a randomized controlled trial. Trials 2009; published 37 Walsh BT. The enigmatic persistence of anorexia nervosa.
online April 23. DOI:10.1186/1745-6215-10-23. Am J Psychiatry 2013; 170: 477–84.
23 Rosenberg WF, Lachin JM. Randomization in clinical trials: theory
and practice. New York: John Wiley and Sons, 2002.

www.thelancet.com Published online October 14, 2013 http://dx.doi.org/10.1016/S0140-6736(13)61746-8 11

 Comment

The challenges of treating anorexia nervosa

The evidence base for anorexia nervosa treatment is For all the psychotherapeutic attention the patients
meagre1–3 considering the extent to which this disorder received, how much progress was made? Body-mass
erodes quality of life and takes far too many lives index (BMI) increased by about 0·70 kg/m² in all three
prematurely.4 But clinical trials for anorexia nervosa are study groups after 10 months of treatment and an

AJ Photo/Science Photo Library

difficult to conduct, attributable partly to some patients’ additional 0·40 kg/m² by 12 months of follow-up.
deep ambivalence about recovery, the challenging task This rise is good because average BMI did not drop
of offering a treatment designed to remove symp­ during the follow-up period and improvements were
toms that patients desperately cling to, the fairly low made on psychological aspects of the disorder. Yet,
prevalence of the disorder, and high dropout rates. at the end of treatment and at 12-month follow-up, Published Online
October 14, 2013
The combination of high dropout and low treatment mean BMI across the three treatment groups was http://dx.doi.org/10.1016/
acceptability has led some researchers to suggest that still in the underweight range. It is noteworthy how S0140-6736(13)61940-6

we pause large-scale clinical trials for anorexia nervosa the bodies (and minds) of individuals with anorexia See Online/Articles
http://dx.doi.org/10.1016/
until we resolve these fundamental obstacles.5,6 nervosa vigorously defend low bodyweight. In terms of S0140-6736(13)61746-8
In The Lancet, Stephan Zipfel and colleagues7 present absolute outcomes, after 10 months of treatment just
results of the Anorexia Nervosa Treatment of OutPatients over a quarter of patients had full syndrome anorexia
(ANTOP) study, in which two manual-based outpatient nervosa (29% receiving focal psychodynamic therapy,
treatments (focal psychodynamic therapy and enhanced 26% assigned enhanced cognitive behaviour therapy,
cognitive behaviour therapy) were compared with and 27% allocated op­timised treatment as usual), and
optimised treatment as usual, which included careful at 12-month follow-up about a fifth of patients had full
and regular monitoring by family doctors linked to care anorexia (21%, 22%, and 28%, respectively). In view of
at specialist treatment centres. The findings provide the generally poor outcomes of anorexia nervosa trials,
some rather sobering observations about treatment we might feel encouraged by these findings. From the
of anorexia nervosa, and highlight the difficulties of outside looking in, we have to do better.
implementing clinical trials for this disorder. We can glean some positive points from the ANTOP
First, ten sites across Germany were needed for recruit­ study. First, the results—and those of a small New Zealand
ment of 242 patients. Second, the trial was facilitated by study8—suggest that treatments that concentrate on
the German health-care system, which provides insurance interpersonal factors might be beneficial in the long term.
coverage for psychotherapy for eating disorders. This Both focal psychodynamic therapy (as presented here) and
system enabled patients assigned to the control arm of interpersonal psychotherapy (done in the New Zealand
optimised treatment as usual to select their preferred study) focus on the role of interpersonal relationships in
type of community intervention. Third, treatment for the maintenance of anorexia nervosa symptoms. Second,
anorexia nervosa takes a long time, at an average of the findings of both studies suggest that variations of
10 months’ duration. No brief interventions for the optimised treatment as usual for anorexia nervosa might
disorder have been judged effective. Fourth, almost be an acceptable therapeutic option. Zipfel and colleagues
a third of patients were lost to follow-up a year after expected their manual-based specialised treatments to
the end of treatment (although, compared with other have the best results, but that expectation wasn’t borne
anorexia trials, this dropout rate is quite good). Fifth, for out, at least in terms of BMI.
many patients, psychotherapy alone did not suffice, and What are the real-world implications of the ANTOP
inpatient treatment was needed for some patients during study findings for clinicians, patients, and their families?
the trial. Allowances should be made for intermittent The results in no way suggest that the only way to
hospitalisations during anorexia outpatient trials because treat anorexia nervosa is via specialised manual-based
recovery from this disorder is rarely linear. Moreover, approaches. This finding is important for two reasons:
precipitous weight loss and other medical complications first, we cannot train every clinician to deliver these
needing hospital treatment do not necessarily mean that specialised treatments; and second, patients might live
outpatient care will ultimately fail. too far from specialist centres to receive manual-based

www.thelancet.com Published online October 14, 2013 http://dx.doi.org/10.1016/S0140-6736(13)61940-6 1

 Comment

treatment delivered by an expert. Results of the ANTOP to chase a societal appearance ideal. Anyone who has
study coupled with the New Zealand findings raise an worked with anorexia patients will attest that even if
important point. In the New Zealand study, specialist the disorder started with a desire to attain a physical
supportive clinical management—designed to be a ideal (and it often does not), the low weight soon eludes
control arm—outperformed both cognitive behaviour wilful control. Identification of aberrant pathways that
therapy and interpersonal therapy in the short term, contribute to these regulatory anomalies and uncovering
and in the ANTOP study, outcomes with optimised gut–brain connections that interrupt the usual processes
treatment as usual did not differ significantly from the of hunger and satiety will hopefully lead to novel ways to
two specialised treatments. Linking family doctors to interrupt this perplexing biological obstinacy.
clinicians skilled in the treatment of eating disorders and
guiding them through the delivery of anorexia nervosa Cynthia M Bulik
treatment while implementing solid principles of clinical University of North Carolina at Chapel Hill, Chapel Hill,
NC 27599, USA
management could be an approach worth considering.
[email protected]
The addition of telemedicine or other technology-aided
CMB is a consultant for Shire Pharmaceuticals.
supervision or consultation might help to extend the
1 Bulik CM, Berkman ND, Brownley KA, Sedway JA, Lohr KN. Anorexia
reach of specialist services. nervosa treatment: a systematic review of randomized controlled trials.
Int J Eat Disord 2007; 40: 310–20.
Despite the positive aspects gleaned from the ANTOP 2 Watson HJ, Bulik CM. Update on the treatment of anorexia nervosa: review
study, we still need to serve patients with anorexia of clinical trials, practice guidelines and emerging interventions.
Psychol Med 2012; published online Dec 10. DOI:10.1017/
nervosa and their families better. We need to discover S0033291712002620.
how to provide better, faster, and lasting results in 3 Hay PPJ, Bacaltchuk J, Byrnes RT, Claudino AM, Ekmejian AA, Yong PY.
Individual psychotherapy in the outpatient treatment of adults with
the management of this disorder. Psychotherapeutic anorexia nervosa. Cochrane Database Syst Rev 2009; published online
Jan 21. DOI:10.1002/14651858.CD003909.
interventions are only partly effective; up to now, 4 Arcelus J, Mitchell AJ, Wales J, Nielsen S. Mortality rates in patients with
no pharmacological agents are effective in the treat­ anorexia nervosa and other eating disorders: a meta-analysis of 36 studies.
Arch Gen Psychiatry 2011; 68: 724–31.
ment of anorexia nervosa. Ongoing work in genetics, 5 Fairburn CG. Evidence-based treatment of anorexia nervosa.
neurobiology, and studies of the microbiome provide Int J Eat Disord 2005; 37 (suppl): S26–30.
6 Halmi K, Agras W, Crow S, et al. Predictors of treatment acceptance and
some hope for the future. We know little about the completion in anorexia nervosa: implications for future study designs.
biological factors that allow individuals with anorexia Arch Gen Psychiatry 2005; 62: 776–81.
7 Zipfel S, Wild B, Groß G, et al, on behalf of the ANTOP study group. Focal
nervosa to defend such low bodyweights and to maintain psychodynamic therapy, cognitive behaviour therapy, and optimised
treatment as usual in outpatients with anorexia nervosa (ANTOP study):
high activity levels in the absence of nutritional fuel. For randomised controlled trial. Lancet 2013; published online Oct 14. http://
too long, people have presumed that anorexia nervosa dx.doi.org/10.1016/S0140-6736(13)61746-8
8 McIntosh V, Jordan J, Carter F, et al. Three psychotherapies for anorexia
is simply a behavioural choice and that all individuals nervosa: a randomized controlled trial. Am J Psychiatry 2005; 162: 741–47.
with the disorder wilfully maintain a low bodyweight

2 www.thelancet.com Published online October 14, 2013 http://dx.doi.org/10.1016/S0140-6736(13)61940-6