REVIEW

Early Response to treatment in Eating Disorders: A Systematic

Review and a Diagnostic Test Accuracy Meta-Analysis
Bruno Palazzo Nazar1,2*, Louise Kathrine Gregor1, Gaia Albano1,3, Angelo Marchica3, Gianluca Lo Coco3,
Valentina Cardi & Janet Treasure1†
1
Department of Psychological Medicine, King’s College London, Institute of Psychiatry, Psychology and Neuroscience (IoPPN), London, UK
2
Federal University of Rio de Janeiro, Institute of Psychiatry (IPUB-UFRJ), Brazil
3
University of Palermo, Department of Psychology and Educational Sciences, Palermo, Italy

Abstract
Objective: Early response to eating disorders treatment is thought to predict a later favourable outcome. A systematic review of the
literature and meta-analyses examined the robustness of this concept.
Method: The criteria used across studies to define early response were summarised following the Preferred Reporting Items for System-
atic Reviews and Meta-Analyses guidelines. Diagnostic Test Accuracy methodology was used to estimate the size of the effect.
Results: Findings from 24 studies were synthesized and data from 14 studies were included in the meta-analysis. In Anorexia Nervosa,
the odds ratio of early response predicting remission was 4.85(95%CI: 2.94–8.01) and the summary Area Under the Curve (AUC) = .77.
In Bulimia Nervosa, the odds ratio was 2.75(95%CI:1.24–6.09) and AUC = .67. For Binge Eating Disorder, the odds ratio was
5.01(95%CI: 3.38–7.42) and AUC = .71.
Conclusion: Early behaviour change accurately predicts later symptom remission for Anorexia Nervosa and Binge Eating Disorder but
there is less predictive accuracy for Bulimia Nervosa. Copyright © 2016 John Wiley & Sons, Ltd and Eating Disorders Association.

Received 19 August 2016; Revised 21 October 2016; Accepted 10 November

2016
Keywords
early response; eating disorders; Anorexia Nervosa; Bulimia Nervosa; binge eating disorder; family therapy; cognitive behavioural therapy

*Correspondence Dr Bruno Palazzo Nazar, Psychological Medicine, King’s College London, Rua Lopes Quintas 100, apt 203 bl2, 22460-010 Rio de Janeiro, Brazil.
Tel: +55 21 3647 3649.
Email:
†
Joint last authors

Published online in Wiley Online Library (wileyonlinelibrary.com) DOI: 10.1002/erv.2495

Introduction purging reduction; early weight gain in AN; early changes to

dietary restriction). This concept is also referred to as ‘rapid’
Time is of the essence in the treatment of eating disorders, as the response (MacDonald, Trottier, McFarlane, & Olmsted, 2015).
longer the duration of untreated illness, the worse the prognosis, An early response to treatment was one of the strongest predictors
particularly for Anorexia Nervosa (AN) (Treasure & Russell, found in a recent meta-analysis of remission at the end of
2011). The response to treatment is greatest in the initial stages treatment and follow-up (Vall & Wade, 2015). Differentiating
of the illness, and it diminishes the longer the disorder persists responders from non-responders early in treatment allows for
(Treasure, Stein, & Maguire, 2015). the possibility of adding potential augmentations to improve
The management of chronic diseases can be enhanced by treat- outcome in those less likely to remit (Madden et al., 2015). This
ment strategies where the treatment plan is modified according to type of design has been introduced in response to family-based
the patient’s history, clinical or psychopathological features treatment (FBT) in adolescent AN (Doyle, Le Grange, Loeb,
and/or their response to previous treatments (Lavori & Dawson, Doyle, & Crosby, 2010; Lock et al., 2015).
2008). This is of great interest for clinicians, as it can offer an op- Interestingly, it seems that early change may predict remission
portunity to personalise therapy in order to optimise outcome across a variety of treatments, regardless of their duration or
(Wilson, Vitousek, & Loeb, 2000). Early and intermediate modality (Fernández-Aranda et al., 2009; Fernàndez-Aranda
markers of treatment response can be used in this individualised et al., 2009).
medicine approach. One limitation of the current literature is the use of many dif-
Early response to eating disorders treatment generally refers to ferent definitions of early response. For example, Bulik, Sullivan,
a clinically meaningful improvement in behavioural symptoms Carter, McIntosh, and Joyce (1999) defined early response to
within the first weeks of treatment (e.g. early binge eating and Bulimia Nervosa (BN) treatment as attainment of remission

Eur. Eat. Disorders Rev. (2016) Copyright © 2016 John Wiley & Sons, Ltd and Eating Disorders Association.
Early Response to Eating Disorder Treatment B. P. Nazar et al.

status at the end of treatment. This is in contrast to many other German languages; if they described criteria used to define early
studies, which have used strategies with statistical modeling to response and remission status. The study selection for screening
find symptom reduction and treatment duration cut-offs. was performed independently by two different authors (GA,
Moreover, the definitions for remission often vary (Agüera AM) and later discussed with two other authors (BPN, VC).
et al., 2013). As yet there are no absolute criteria for either the Two authors (GA, LG) extracted data under the supervision of
timing considered as ‘early’ or the amount of symptom change the first author (BPN). All selection procedures are demonstrated
used to demarcate an early response. Also, these criteria may vary in Figure 1.
depending on the population studied, treatment plan and setting.
A recent review and meta-analysis of 34 articles on the rapid Meta-analysis
response to eating disorder treatment (Linardon, Brennan, & The statistical methods developed for meta-analysis of diagnostic
de la Piedad Garcia, 2016) synthesized the size of the effect of test accuracy were employed in order to test the predictive validity
various definitions of ‘early response’ to predict remission of the early response/remission paradigm. A participant was con-
(Linardon et al., 2016). They defined early response in terms of sidered positive if remission was achieved. In line with this, early
any change in eating disorder symptomatology within the first response subjects achieving remission were considered true posi-
half of the treatment delivered. The authors found that early tives; non-early responders that achieved remission were the false
response was associated with sustained end of treatment and positives; non-early responders that did not achieve remission
follow-up improvements in eating disorder symptoms, with no were the true negatives and the early response subjects that did
moderator effects such as type of eating disorder diagnosis, not remit were the false negatives.
treatment modality and the criteria used to define an early The accuracy measures reported by a diagnostic test accuracy
response. The definition of ‘early’ and the length of treatment meta-analysis include sensitivity, specificity, positive likelihood
were not examined because of large variability. Also, they did ratios and negative likelihood ratios. Sensitivity refers to how well
not provide a quantitative synthesis of early weight gain in AN a test can predict a given outcome when this is present (e.g. presence
or a qualitative synthesis of BN and BED results. Thus, the of disease), while specificity points to the capacity of a test to rule
concept of early response still requires further investigation and out an outcome when it is really absent. The likelihood ratios are
clarification. One method to do this may be to calculate diagnos- used to indicate the probability of a diagnosis using the presence
tic test accuracy measures for each eating disorder syndrome. of a marker (Cochrane handbook for DTA reviews, 2010).
Diagnostic test accuracy methodology is routinely used to Likelihood ratios range from zero to infinity, with larger ratios
determine reliable cut-off points to predict a disease state demonstrating increased probability that a sign or test indicates a
(Macaskill, Gatsonis, Deeks, Harbord, & Takwoingi, 2010). In positive diagnosis; ratios equal to 1 lack any diagnostic value and
mental health studies (Takwoingi, Riley, & Deeks, 2015), it has ratios from 0 to 1 indicate absence of the diagnosis (Attia, 2003).
been used to establish the reliability of screening and diagnostic There is a greater chance of obtaining the investigated outcome as
measures (Mitchell, Shukla, Ajumal, Stubbs, & Tahir, 2014; Manea, the positive likelihood ratio increases. Conversely, it is less likely
Gilbody, & McMillan, 2015). We have employed this methodology to find the outcome as the negative likelihood ratio decreases. The
to determine the accuracy of predicting end of treatment last accuracy measure is the Diagnostic Odds Ratio (DOR), which
remission status from the early response to treatment for AN, is an indicator of test performance providing a measure of the
BN and Binge Eating Disorder (BED). strength of association between a test and an outcome.
The aim of this systematic review is to collate the literature Additionally, the diagnostic test accuracy meta-analysis can
relating to the early response to treatment in eating disorders combine the different Receiver Operating Characteristic (ROC)
and conduct a meta-analysis using diagnostic test accuracy curves from each study to produce a synthetic curve, named Sum-
methodology to examine the robustness of the early response mary Receiver Operating Characteristic (SROC). Using the SROC
concept as a predictor of outcome. curve, it is possible to produce a summary Area Under the Curve
(AUC) measure. This method takes into account the rate of false
Methods positives and true positives at different cut-off values, produced
by sensitivity and specificity of a test for an outcome. The best test
Systematic review has an AUC of 1, while a poor test has an AUC of 0.5 (Lalkhen &
We performed a systematic review following Preferred Reporting McCluskey, 2008).
Items for Systematic Reviews and Meta-Analyses (PRISMA) guide- In order to include papers in the meta-analysis, we needed data
lines (Moher, Liberati, Tetzlaff, & Altman, 2009) using the PubMed, on the true positives, false positives, true negatives and false
PsychInfo, Embase, Scopus and Web of Knowledge databases, with negatives for any given categorisation of early response and the
the following search terms: (‘earl* response’ OR ‘short-term number of patients achieving remission. Thus, we only included
response’ OR ‘rapid response’) AND (‘Eating disorder’ OR studies that provided information about the number of partici-
‘Anorexia Nervosa’ OR ‘Bulimia Nervosa’ OR ‘Binge Eating’ OR pants that achieved remission in the early response and non-early
‘EDNOS’). Subsequently, a hand-search of the reference lists from response groups, as well as the total number of participants
selected papers was analysed, and authors were contacted when classified as early responders and the number of participants that
further information was not reported in the manuscript. achieved remission. The accuracy measures were calculated from
Articles were considered eligible for inclusion in the review if the pooled values for sensitivity, specificity, positive and negative
they were published any time before August 2016 (week 4); if they likelihood ratios and DOR, all obtained using the Meta-Disc pack-
were published in English, Spanish, Portuguese, French or age (Zamora, Abraira, Muriel, Khan, & Coomarasamy, 2006).

Eur. Eat. Disorders Rev. (2016) Copyright © 2016 John Wiley & Sons, Ltd and Eating Disorders Association.
B. P. Nazar et al. Early Response to Eating Disorder Treatment

Figure 1. Search flow diagram for research about Early Response in the treatment of eating disorders

Further information, not reported in the published manu- Gueorguieva, & Masheb, 2012; Grilo, Masheb, & Wilson, 2006;
script, was kindly obtained directly from the main authors of Grilo & Masheb, 2007; Grilo, White, Masheb, & Gueorguieva,
Schlup, Meyer, and Munsch (2010), Vaz, Conceição, and 2015; Masheb & Grilo, 2007; Safer & Joyce, 2011; Schlup et al.,
Machado (2014), Le Grange, Accurso, Lock, Agras, and Bryson 2010) were included in the BED meta-analysis.
(2014) and MacDonald, Trottier, McFarlane, and Olmsted The demographic characteristics of participants, study design
(2015) in order to complete the meta-analysis. and treatment administered during the original trials are pre-
sented in Table 1. The definition of early response varied primar-
Publication bias ily according to the eating disorder diagnosis of interest. The main
Risk of bias across studies was examined with Q and I2 statistics, outcome for AN was the attainment of 95% of Ideal Body Weight
but because they do not account for specific diagnostic test accu- (IBW), whereas for BN, it was a percentage reduction in bingeing
racy biases, such as threshold effects, a scatterplot of studies in the and/or purging over a variable time period. For BED it was a per-
SROC graphic, was examined (Macaskill et al., 2010). Each indi- centage reduction in bingeing over a period of time, and in some
vidual study is represented as a point in space across the SROC studies weight loss was also included as an outcome. The IBW
graphic, and bias can be analysed through visual inspection. measure used across studies was based on Center for Disease
Control growth charts that are routinely used in clinical settings,
Results with cut-offs defined through epidemiological studies
(Kuczmarski et al., 2000). The only study that did not state how
The search resulted in a final selection of 24 articles for the qual- they defined the percentage of IBW was Lock, Couturier, Bryson,
itative synthesis. For the meta-analysis of AN results, only three and Agras, 2006. Some studies required improvements in psycho-
studies (Le Grange et al., 2014; Madden et al., 2015; Wales et al., pathology, measured by the EDE-Q, to fall within 1 (Doyle et al.,
2016) had the variables of interest. Two of these studies (Le 2010; Le Grange et al., 2014; Madden et al., 2015) or 2 (Lock et al.,
Grange et al., 2014; Madden et al., 2015) performed two separate 2006) standard deviations of community norms in combination
analyses using different definitions of early response, which were with weight gain to define AN remission,
both examined. For BN, four studies (MacDonald et al., 2015; The studies either employed a hypothesis generating approach,
Raykos et al., 2014; Thompson-Brenner, Shingleton, Sauer- where they selected the time point for defining early response
Zavala, Richards, & Pratt, 2015; Vaz et al., 2014) had information from their statistical analysis, or used a hypothesis testing
available for synthesis, while seven studies (Grilo, White, Wilson, approach, whereby criteria used in previous studies were used to

Eur. Eat. Disorders Rev. (2016) Copyright © 2016 John Wiley & Sons, Ltd and Eating Disorders Association.
 Table 1 Summary of demographic and clinical characteristics of included studies about Early Response in the treatment of Eating Disorders

Mean BMI of
Sample Mean age of total sample at Diagnoses Duration Control
Author Year size Sex total sample (SD) baseline (SD) of sample Study design of intervention Treatment condition

Agras 2000 194 194 F 28.1 (+7.9) N.R. BN Observational 18 sessions over 16 weeks Cognitive Behavioural Therapy N.A.
et al. before-after study (twice weekly for the first
2 weeks, weekly thereafter)
Doyle 2010 65 58 F 7 M 14.9 (+2.1) 17.0 (+1.7) AN Observational 20 sessions over 12 months Family-Based Treatment N.A
et al. before-after study
Fairburn et al. 2004 220 220 F N.R. N.R. BN Randomised 19 sessions over a period Cognitive Behavioural Therapy Interpersonal Psychotherapy
controlled trial of 20 weeks
Grilo 2014 104 73 F 31 43.9 (+11.2) 38.3 (+5.6) BED Randomised 16 weeks (sessions per Self-Help Cognitive Sibutramine
et al. M controlled trial week N.R.) Behavioural Placebo
Early Response to Eating Disorder Treatment

Treatment + Sibutramine Self-Help Cognitive

Behavioural Therapy + Placebo
Grilo 2012 90 56 F34 44.89(+9.48) 38.65(+5.70) BED Randomised 16 group sessions Cognitive Behavioural Therapy Behavioural Weight Loss
et al. M controlled trial over 24 weeks
Grilo 2007 50 44 F6 M 47.0 (+7.0) 36.0 (+4.7) BED Randomised 6 sessions over 12 weeks Cognitive Behavioural Cognitive Behavioural
et al. controlled trial Therapy Guided Therapy Guided
Self-Help + Orlistat Self-Help + Placebo
Grilo 2006 108 84 F24 44.0(+8.6) 36.3 (+7.9) BED Randomised 16 sessions over Cognitive Behavioural Cognitive Behavioural
et al. M controlled trial 16 weeks Therapy + Fluoxetine Therapy + Placebo
Hartmann 2007 85 N.R. 25.05(+6.51) 13.88 (+1.34) AN Observational N.R. Inpatient Treatment N.A.
et al. before-after study
Hilbert et al. 2015 205 N.R. N.R. N.R. BED Randomised 16 sessions over Interpersonal Psychotherapy Self-Help Cognitive
controlled trial 16 weeks followed by Behavioural Therapy
4 fortnightly sessions Behavioural Weight Loss
Le Grange 2014 121 110 F11 14.4 (+1.6) 16.1 (+1.1) AN Randomised 24 sessions over Family-Based Treatment Adolescent
et al. M controlled trial 12 months Focused Therapy
Le Grange 2008 80 78 F2 M 16.1(+1.6) 22.1 (+3.0) BN Randomised 20 sessions over 6 months Family-Based Therapy Individual
et al. controlled trial Supportive Psychotherapy
Lock 2006 86 78 F8 M 15.2 (+1.7) 16.0 (+1.6) AN Randomised 10 sessions over 6 months Family-Based Family-Based
et al. long-term at identification controlled trial (short-term) or Therapy Short-term Therapy Long-term
family-based 17.1 (+1.4) 20 sessions over (20 sessions
therapy (n = 42) at randomisation 12 months (long-term) 12 months)
15.2 (+1.6)
short-term
family-based
therapy (n = 44)
MacDonald 2015 158 152 F6 27.1 (+8.8) 23.2 (+4.4) BN Observational Open agenda; up to Day Hospital Treatment N.A.
et al. M before-after study 35–40 h per week
(sessions per week
N.A. as continuous)
Madden 2015 82 78F4M 14.67 (+1.41) N.R. AN Randomised 20 sessions outpatient FBT Long-stay Hospitalisation Short-stay Hospitalisation
et al. controlled trial for Weight restoration for Medical Stabilisation
Marrone 2009 116 114 F2 N.R. N.R. BN Randomised Telemedicine Cognitive Face-to-Face Cognitive
et al. M controlled trial Behavioural Therapy Behavioural Therapy

Continues
B. P. Nazar et al.

Eur. Eat. Disorders Rev. (2016) Copyright © 2016 John Wiley & Sons, Ltd and Eating Disorders Association.
 B. P. Nazar et al.

Table 1. (Continued)

Mean BMI of
Sample Mean age of total sample at Diagnoses Duration Control
Author Year size Sex total sample (SD) baseline (SD) of sample Study design of intervention Treatment condition
20 sessions over 16 weeks
(twice weekly for first
fortnight, weekly thereafter)
Masheb 2007 75 61 F14 46.0 (+9.1) 35.3 (+6.9) BED Randomised 6 sessions over Cognitive Behavioural Behavioural Weight
et al. M controlled trial 12 weeks + self-help Therapy Guided Loss Guided Self-Help
Self-Help
McFarlane 2008 58 58 F 29.8 (+9.4) 23.2 (+7.3) 18 = AN16 = BN Observational N.R. Day Hospital Treatment N.A.
et al. 24 = EDNOS before-after study
Raykos 2013 105 N.R. 25.9 (+8.9) N.R. 17 = AN50 = BN Observational 20 sessions Enhanced Cognitive N.A.
et al. 38 = EDNOS before-after study over 20 weeks Behavioural Therapy
Safer 2011 101 86 F15 52.1 (+10.6) N.R. BED Randomised 20 sessions Dialectical Behaviour Therapy Active Comparison
et al. M controlled trial over 20 weeks Group Therapy
Schlup 2010 76 76 F 44.5(+10.7) 33.2 (+5.6) BED Non-randomized 16 sessions over Cognitive Behavioural Therapy Cognitive Behavioural
et al. controlled trial 16 weeks then 6 sessions Long-term and Short-term Therapy Short-term (8 sessions)
over 6 months
(long-term) or
8 sessions over 8 weeks

Eur. Eat. Disorders Rev. (2016) Copyright © 2016 John Wiley & Sons, Ltd and Eating Disorders Association.
then 5 sessions over
12 months (short-term)
Thompson-Brenner2015 43 43 F 25.7 (+8.4) 23.5 (+3.5) BN Randomised 2 sessions per week Enhanced Cognitive N.A.
et al. controlled trial over 4 weeks Behavioural Therapy
Vaz 2014 42 42 F 26.3(+7.02) 22.7 (+4.0) BN Observational 1 fortnightly session Self-Help Cognitive Behavioural N.A.
et al. before-after study over 16 weeks Therapy + Group Meetings
Wales 2015 102 82F20M 26.56 (+8.95) 13.07 (+1.42) AN Observational Open agenda, Inpatient Treatment, Weight N.A.
et al. before-after study (sessions per week Restoration + Cognitive
N.A. as continuous) Behavioural Therapy + Group Therapy
Zunker 2010 179 161 F18 46.5(+10.2) 39.25 (N.R.) BED Randomised 15 sessions Self-Help Cognitive Therapist-led
et al. M controlled trial over 20 weeks Behavioural Therapy + Group MeetingsCognitive Behavioural Therapy
Therapist-assisted
Cognitive Behavioural Therapy

Legend: N.A. = not applicable; N.R. = not reported; F = female; M = male.

Early Response to Eating Disorder Treatment
 Table 2 Summary of statistical modeling and definitions of Early Response and Remission in the included studies

Patients within ER category Definition of Remission (percentage

Author, year Patient group Statistical modelling for defining ER ER measure and time point (percentage and number) and number of remitted patients)

Doyle, 2010 Adolescent ROC curve and AUC analysis Weight gain of 1.61% at session 3 (% and n N.R.) Achieving 95% of IBW at EOT
outpatient and 2.68% at session 4 predicts 47.7% (n = 31) of patients achieved
weight gain at EOT remission
2
Hartmann, 2007 Adolescent Growth curve analysis; non-linear Weight gain of 0.46 BMI points in (% and n N.R.) Achieving BMI of 17.5 kg/m or
inpatient regression; ROC curves weeks 1–2 and 0.38 BMI points in gaining 2 kg
weeks 3–4 predicts remission at EOT (% and n N.R.)
Le Grange, 2014 Adolescent ROC curves and AUC analysis; t-test Weight gain on any session from 3 13.1% (n = 7) of patients Achieving 95% of IBW at EOT +
outpatient and Chi-square tests to 8 in FBT or 4 to 6 in AFT at the strongest predictor EDE-Q within 1 S.D. of norm
predicts weight gain at EOT of FBT (session 8) and 42% (n = 21) of patients in FBT and
35% (n = 19) of patients 23% (n = 12) of patients in AFT
Early Response to Eating Disorder Treatment

at the strongest predictor achieved remission

of AFT (session 5)
showed ER.
Lock, 2006 Adolescent Intention-to-treat analysis; ROC Weight gain of 2.3 (+1.9) kg by (% and n N.R.) Achieving 95% of IBW + EDE-Q
outpatient curves and AUC analysis; logistic session 2 within 2 S.D. of norm at EOT
regression 4 (+3.0) kg by session 9; 4.35 (+2.8) 62% (n = 42) of patients achieved
kg by session 10 predicts remission
weight gain at EOT
Madden, 2015 Adolescent ROC curve analysis; t-test and Chi- Weight gain of 1.8 kg at session 4 35% (n = 24) of patients Strict definition: Achieving 95% of IBW
outpatient square tests; logistic regression predicts remission at EOT showed ER at EOT + EDE-Q within 1 S.D. of norm
and inpatient 18.5% (n = 16) of patients achieved
strict remission
Broad definition: 95% EBW at EOT +
EDE-Q within 2 S.D. of norm
McFarlane, 2008 Adult Kaplan–Meier survival analysis 90% adherence to prescribed meal 69% of entire Partial remission definition: no more
outpatient plan for at least 2 weeks within first Transdiagnostic sample than 2 bingeing and/or purging episodes
3 weeks showed ER. ER N.R. by per month for 2 months, achieving and
diagnostic category maintain a BMI of 19.5 or higher for
two months, and adhering to a
prescribed normalised meal plan for
two months
48% (n = 8) of AN sample achieved
remission
Raykos, 2013 Adolescent and Chi-square tests; ANOVA A reduction on the EDE-Q global by 11% (n = 4) of patients A post-treatment BMI >18.5
adult outpatient at least 1.52 points in the first 3–6 showed ER A global EDE-Q score within 1 S.D.
weeks of norm
Abstinence from binge eating and
purging
for 28 days prior to the end of treatment

41% (n = 7) of patients achieved full

remission; 29% (n = 5) of patients
achieved partial remission
Wales, 2015 Adult inpatient Logistic regression analysis

Continues
B. P. Nazar et al.

Eur. Eat. Disorders Rev. (2016) Copyright © 2016 John Wiley & Sons, Ltd and Eating Disorders Association.
 Table 2. (Continued)

Patients within ER category Definition of Remission (percentage

Author, year Patient group Statistical modelling for defining ER ER measure and time point (percentage and number) and number of remitted patients)
2
Weight gain of 0.5—1 kg per week 51.7% (n = 45) of patients Achieving a BMI of 17.5 kg/m
B. P. Nazar et al.

during first 6 weeks predicts showed ER 82.2% (n = 37) of ER and 35.7%

remission at EOT (n = 15) of NER achieved remission
Fairburn, 2004 Adult outpatient Results by Agras et al. (2000) to Reduction of >49% in purging after (% and n N.R.) Abstinence from bingeing and purging
define ER as same group 4 weeks (session 6) predicts remission in last 28 days prior to EOT
17.7% (n = 39) of patients achieved
remission
Le Grange, 2008 Adolescent ROC curves and AUC analysis Reduction of >85% in bingeing and (% and n N.R.) Abstinence from bingeing and purging
outpatient purging after 6 weeks (session 6) at EOT
predicts remission (% and n N.R.)
MacDonald, 2015 Adult day ROC curves and AUC analysis Using frequency criteria: <3 in the first Using frequency criteria: < 1 bingeing and/or purging episode
hospital 4 weeks or <1 in the first 2 weeks 65.8% and 61.4% in the last 2 weeks of Day Hospital
Using percentage criteria: >99.7% in Using percentage criteria: and < 1 episode in the 1st month after
the first 4 weeks or >95.7% in the first 32.3% and 51.9% Day Hospital ended
2 weeks 79% (n = 76) of patients with ER
achieved remission; 41.6% (n = 25) of
patients with NER achieved remission
Marrone, 2009 Adult outpatient ROC curves and AUC analysis 44.51% reduction in bingeing at week (% and n N.R.) Abstinence from bingeing and purging
6 for telemedicine CBT and 87.21% in the last 28 days prior EOT
reduction in bingeing at week 8 for 38.8% (n = 45) of patients achieved
face-to-face CBT remission
McFarlane, 2008 Adult outpatient Kaplan–Meier survival analysis 90% adherence to prescribed meal 69% of entire Partial remission definition: no more
plan for at least two weeks within the transdiagnostic sample – than 2 bingeing and/or purging episodes
first three weeks of treatment not clear how many of per month for 2 months, achieving and

Eur. Eat. Disorders Rev. (2016) Copyright © 2016 John Wiley & Sons, Ltd and Eating Disorders Association.
this had BN maintain a BMI of 19.5 or higher for
two months, and adhering to a
prescribed normalised meal plan for
two months
59% (n = 11) of BN sample achieved
remission

Thompson- Adult outpatient ROC curves and AUC analysis 65% reduction in purging or 25% For purging, 44.2% Abstinence from bingeing and purging
Brenner, 2015 reduction of BDI by week 4 (n = 19) of patients at EOT
showed ER. For BDI, 44.2% (n = 19) of patients achieved
53.4% (n = 23) of remission
patients showed ER 79% (n = 15) of patients with ER and
29% (n = 7) of patients with NER
achieved remission
Vaz, 2014 Adult outpatient Intention-to-treat analysis; logistic Arbitrary, selecting 8-month FU 50% (n = 21) of patients Abstinence from bingeing and purging
regression; survival analysis percentage (51% reduction) from showed ER 4 weeks prior EOT
Fairburn (2004) as criterion, 52.3% (n = 22) of patients achieved
expanding to binge symptoms as remission
well; at week 3 28.6% (n = 6) of patients with ER and
38% (n = 8) of patients with NER
achieved remission

Continues
Early Response to Eating Disorder Treatment
 Table 2. (Continued)

Patients within ER category Definition of Remission (percentage

Author, year Patient group Statistical modelling for defining ER ER measure and time point (percentage and number) and number of remitted patients)

Grilo, 2006 Adult outpatient ROC curves and AUC analysis >65% reduction in binge eating 44% (n = 48) of patients Abstinence from bingeing in 28 days
episodes by week 4 predicts remission showed ER prior to EOT
at EOT 40.7% (n = 44) of patients achieved
remission
60.4% (n = 29) of patients with ER and
25% (n = 15) of patients with NER
achieved remission
Grilo, 2007 Adult outpatient ROC curves and AUC analysis >70% reduction in binge eating 42% (n = 21) of patients Zero binges (OBEs) during the previous
episodes by week 4 predicts remission showed ER month determined by EDE interview
at EOT 76.2% (n = 16) of patients with ER and
Early Response to Eating Disorder Treatment

31% (n = 9) of patients with NER

achieved remission
Grilo, 2012 Adult outpatient ROC curves and AUC analysis >70% reduction in binge eating 67% (n = 30) of patients Zero binges (OBEs) during the previous
episodes by week 4 predicts remission in CBT and 47% (n = 21) 28 days determined by EDE interview
at EOT of patients in BWL 58.8% (n = 30) of patients with ER and
showed ER 17.9% (n = 7) of patients with NER
56.7% (n = 51) in total achieved remission
showed ER
Grilo, 2014 Adult outpatient Definition of ER based on Grillo et al. >65% reduction in binge eating 47% (n = 49) patients Zero binges (OBEs) during the previous
(2006), which obtained definition by ROC episodes by week 4 predicts remission showed ER 28 days determined by EDE interview
and AUC. at EOT 51% (n = 25) of patients with ER and 9%
(n = 5) of patients with NER achieved
remission.
Hilbert, 2015 Adult outpatient Definition of ER based on Grilo et al. >70% reduction in binge eating 70.7% (n = 145) of study Zero binges (OBEs) during the previous
(2012). This study obtained definition by episodes by week 4 predicts remission patients showed ER 28 days determined by EDE interview
ROC and AUC. at EOT 73.4% (n = 47) patients in (% and n N.R.)
BWL; 74.2% (n = 49)
patients in CBT and 65.3%
(n = 49) patients in IPT
showed ER
Masheb, 2007 Adult outpatient ROC curves and AUC analysis; maximum >65% reduction in binge eating 54.7% (n = 41) of patients Abstinence from bingeing in the last 28
likelihood linear mixed model analysis; Chi episodes at week 4 predicts remission showed ER days prior to EOT.
square analysis at EOT 46.3% (n = 19) of patients with ER and
14.7% (n = 5) with NER achieved
remission
Safer, 2011 Adult outpatient Chi square analysis ≥ 65% reduction in frequency of binge 40.6% (n=41) of patients Abstinence from binge eating, defined as
eating episodes at week 4 predicts showed ER no OBE days (per the EDE) over the 28
remission at EOT following Grilo et al. days prior to EOT.
(2006) and Masheb and Grilo (2007). 70.7% (n = 29) patients with ER and
33.3% (n = 20) of patients with NER
achieved remission
Schlup, 2010 Adult outpatient Based on definition provided by Grillo and >65% reduction in binge eating 55% (n = 42) of patients Abstinence from bingeing in the last 28
Masheb. (2007), Masheb and Grillo. (2007) episodes at week 4 predicts remission showed ER days prior to EOT.
and Grillo, Masheb and Wilson. (2006). at EOT 51% (n = 39) of patients achieved
remission
B. P. Nazar et al.

Eur. Eat. Disorders Rev. (2016) Copyright © 2016 John Wiley & Sons, Ltd and Eating Disorders Association.
Continues
B. P. Nazar et al. Early Response to Eating Disorder Treatment

Legend: N.A. = not applicable; N.R. = not reported; EOT = end of treatment; ER = early response; NER = non-early response; BMI = body mass index; IBW = ideal body weight; EBW = expected body weight; EDE-
assign groups. The definitions for response and remission

Definition of Remission (percentage

and number of remitted patients)

50% (n = 21) of patients with ER and

provided by each study, as well as the statistical modeling

41.2% (n = 14) of patients with NER

Abstinence from binge eating for 30

employed to obtain them and the AUC yielded by such methods,
are presented in Table 2. The different time points that defined an
‘early’ response in each study are also presented in Table 2.
Early response for Anorexia Nervosa
achieved remission

days prior to EOT

(% and n N.R.)
Although only three studies could be included in the meta analysis,
eight studies (Doyle et al., 2010; Hartmann, Wirth, & Zeeck, 2007;
Le Grange et al., 2014; Lock et al., 2006; MacDonald et al., 2015;
Madden et al., 2015; Raykos, Watson, Fursland, Byrne, & Nathan,
2013; Wales et al., 2016) included patients with AN. The character-
istics of all AN studies are summarized in Tables 1 and 2.
Patients within ER category

73.7% (n = 141) of patients

(percentage and number)

Early weight gains in outpatient adolescents with AN were

predictive of remission in studies using FBT (Doyle et al., 2010; Le
Grange et al., 2014; Lock et al., 2006), adolescent focused therapy
(Le Grange et al., 2014) and cognitive behavioural therapy adapted
to eating disorders (CBT-E) (Raykos et al., 2013). The same result
showed ER

was found with inpatient AN, where early weight gains at different
time points during the first six weeks of treatment could predict
remission (Hartmann et al., 2007; Wales et al., 2016).
15% reduction in binge eating episodes

Early response studies in Bulimia Nervosa

by week 1 predicts remission at EOT

Altogether nine studies investigated the predictive value of early

ER measure and time point

response to BN treatment, two of which were included in

transdiagnostic samples described above (McFarlane et al., 2008;
Raykos et al., 2013). Eight studies assessed adult outpatients (Agras
et al., 2000; Fairburn, Agras, Walsh, Wilson, & Stice, 2004;
Marrone, Mitchell, Crosby, Wonderlich, & Jollie-Trottier, 2009;
Thompson-Brenner et al., 2015; Vaz et al., 2014), one study
reported results for adolescent outpatients (Le Grange, Doyle,
Crosby, & Chen, 2008), one investigated adolescent and adult
outpatients (Raykos et al., 2013) and two analysed adult day
hospital treatment (MacDonald et al., 2015; McFarlane et al., 2008).
All interventions consisted of psychotherapy without medication.
ROC curves and AUC analysis; intention-
These definitions were obtained through
Statistical modelling for defining ER

Interestingly, the predictive utility of early response has been

demonstrated across different forms of cognitive behaviour
therapy (CBT) treatment delivery, either face-to-face (Agras et al.,
ROC curve and AUC analysis.

2000; Agüera et al., 2013; Vaz et al., 2014), through telemedicine

(Marrone et al., 2009) or in self-help formats (either guided or
not) (Vaz et al., 2014). Other studies also found that early
response to interpersonal therapy (Fairburn et al., 2004) in adult
to-treat analysis

BN, and to either individual supportive psychotherapy or FBT (Le

Grange et al., 2008) in adolescent BN, could predict remission. It
has been argued that longer protocols are needed, as brief
Q = Eating Disorders Examination Questionnaire.

psychotherapeutic interventions did not find early responders to

achieve remission more frequently (Fernàndez-Aranda et al., 2009).
Reductions in frequency or percentage of binge/purge episodes
Adult outpatient
Patient group

were also associated with end of treatment remission in adult BN

patients receiving day hospital treatment (MacDonald et al., 2015).
Not only is an early reduction in bulimic symptoms indicative
of later remission, but reductions in depressive symptoms early in
Table 2. (Continued)

treatment are also predictive of treatment outcome in BN

(Thompson-Brenner et al., 2015).
Zunker, 2010
Author, year

Early response studies in Binge Eating Disorder

Nine studies investigated whether an early reduction in binge
eating predicted remission at discharge from outpatient treatment

Eur. Eat. Disorders Rev. (2016) Copyright © 2016 John Wiley & Sons, Ltd and Eating Disorders Association.
Early Response to Eating Disorder Treatment B. P. Nazar et al.

in adults with BED (Grilo et al., 2012; Grilo et al., 2006; Grilo & The initial meta-analysis conducted on AN studies included
Masheb, 2007; Grilo et al., 2015; Hilbert, Hildebrandt, Agras, data from Le Grange et al. (2014), Madden et al. (2015) and Wales
Wilfley, & Wilson, 2015; Masheb & Grilo, 2007; Safer & Joyce, et al. (2016). The data from AN studies (n = 3) (Le Grange et al.,
2011; Schlup et al., 2010; Zunker et al., 2010). Each of these studies 2014; Madden et al., 2015; Wales et al., 2016), using results from
examined the early response to psychological treatments, while week four, provided a pooled sensitivity of 0.57 (95% CI: .49–.66)
three also included an adjunctive pharmacological treatment and a specificity of 0.78 (95% CI: .72–.84). The pooled positive
(Grilo et al., 2006; Grilo & Masheb, 2007; Grilo et al., 2015). Three likelihood ratio was 2.43 (95%CI: 1.82–3.25), while the negative
studies (Grilo et al., 2015; Hilbert et al., 2015; Schlup et al., 2010) likelihood ratio was 0.58 (95% CI: .42–.81). The summary DOR
used a hypothesis testing approach, while the others used a was 4.85 (95% CI: 2.94–8.01). AUC analysis revealed a pooled
hypothesis generating approach. For several definitions of early
response, Zunker et al. (2010) reported that, only binge reductions
at week one predicted remission in participants receiving CBT.
For both CBT and behavioural weight loss interventions, Grilo
et al. (2012) and Masheb and Grilo (2007) found that early reduc-
tions in binge episode frequency predicted remission at end of
treatment. Hilbert et al. (2015) found similar results for CBT in
the guided self-help format but not for interpersonal therapy or
behavioural weight loss treatment.
Analysing studies using medication associated with psychother-
apy, Grilo et al. (2006) found that binge reductions by week 4
predicted remission in a fluoxetine plus CBT trial. Studies using
CBT combined with anti-obesity agents found similar results
(Grilo & Masheb, 2007; Grilo et al., 2015).

Quantitative synthesis (meta-analysis)

The results from the meta-analysis are presented in Table 3. The
synthesised graphics for AN, BN and BED accuracy measures are
presented in the online Supporting Information. The SROC
curve for BED is presented as an example in Figure 2, and the
ones for AN and BN are in the online Supporting Information. Figure 2. Synthetic receiver operating characteristics curve for BED

Table 3 Summary of diagnostic test accuracy meta-analyses

Q-statistics
# 2 2
Group analysed (sample size) Accuracy measure Pooled value (95% CI) Interpretation X p-Value I

AN (n = 3) Sensitivity .57 (.49–.66) High specificity/low sensitivity 21.61 <.001 81.5%

Specificity .78 (.72–.84) High specificity/low sensitivity 8.76 .06 54.4%
Positive likelihood ratio 2.43 (1.82–3.25) +(15–30)% 1.72 .78 0%
Negative likelihood ratio .58 (.42–.81) (0–15)% 11.47 .02 65.1%
Diagnostic odds ratio 4.85 (2.94–8.01) N.A. 3.29 .51 0%
AUC (S.E.) .77 (.03)* Moderate accuracy N.A. N.A. N.A.
BN (n = 4) Sensitivity 0.64 (0.56–0.71) Balanced specificity/sensitivity 20.91 <.001 85.7%
Specificity 0.63 (0.54–0.71) Balanced specificity/sensitivity 3.88 .27 22.6%
Positive likelihood ratio 1.67 (1.19–2.33) + (0–15)% 4.11 .25 27%
Negative likelihood ratio 0.62 (0.37–1.02) (0–15)% 13.03 .005 77%
Diagnostic odds ratio 2.75 (1.24–6.09) N.A. 6.42 .09 53.3%
AUC (S.E.) 0.67 (0.04) Low accuracy N.A. N.A. N.A.
BED (n = 7) Sensitivity 0.59 (0.54–0.65) High specificity/low sensitivity 10.06 .12 40.3%
Specificity 0.75 (0.69–0.79) High specificity/low sensitivity 25.85 <.001 76.8%
Positive likelihood ratio 2.41 (1.69–3.43) +(15–30)% 15.54 .01 61.4%
Negative likelihood ratio 0.53 (0.46–0.68) (0–15)% 3.81 .70 0%
Diagnostic odds ratio 5.01 (3.38–7.42) N.A. 6.38 .38 5.9%
AUC (S.E.) 0.71 (0.03) Moderate accuracy N.A. N.A. N.A.

Legend: AUC = area under the curve; S.E. = standard error; N.A. = not applicable.
*Pooled data excluding 1 outlier (Raykos).
#
For positive likelihood ratio and negative likelihood ratio, refer to approximate % of increase or decrease in presence of outcome; For AUC, refer to diagnostic accuracy of a
test identifying outcome.

Eur. Eat. Disorders Rev. (2016) Copyright © 2016 John Wiley & Sons, Ltd and Eating Disorders Association.
B. P. Nazar et al. Early Response to Eating Disorder Treatment

area of 0.77 (S.E. = .03). This value suggests that the early response of remission. However, the duration of abstinence required in
is a good predictor of AN remission. order to be classified as remitted varied across studies; while most
For BN (n = 4) (MacDonald et al., 2015; Raykos et al., 2013; studies used a four week period, others used one or two week
Thompson-Brenner et al., 2015; Vaz et al., 2014), only adult abstinence as the criteria (Williams, Watts, & Wade, 2012). This
studies had data available for the meta-analysis. The summary of variation may account for the low accuracy. The small number
the combined sensitivity was of 0.64 (95% CI: 0.56–0.71), while of studies precluded the use of a meta-regression.
the specificity was 0.63 (95% CI: 0.54–0.71). The positive The results of the meta-analysis for the BED studies demon-
likelihood ratio for BN studies was 1.67 (95% CI: 1.19–2.33), strated a high specificity and low sensitivity in predicting remis-
and the negative likelihood ratio was 0.62 (95% CI: 0.37–1.02). sion. There was an approximate 30% increase in the chance of
The pooled DOR was 2.75 (95% CI: 1.24–6.09). The combined remission in a BED patient with an early response based on the
AUC was 0.67 (S.E. = 0.04), which suggests that it has a moderate positive likelihood ratio. Using pooled DOR, BED patients with
capacity for predicting remission. an early response were five times more likely to achieve remission.
In the BED studies (n = 7) (Grilo et al., 2012; Grilo et al., 2006;
Grilo & Masheb, 2007; Grilo et al., 2015; Masheb & Grilo, 2007; Strengths and limitations
Safer & Joyce, 2011; Schlup et al., 2010), the pooled sensitivity
There were too few studies available for a meta-regression to
was 0.59 (95% CI: 0.54–0.65), with a specificity of 0.75 (95%
examine factors that might explain the variance. Access to
CI: 0.69–0.79). The positive likelihood ratio for BED studies was
individual data would enable the different early response and
2.41 (95% CI: 1.69–3.43), and the negative likelihood ratio was
remission criteria to be modeled. Furthermore, although the early
0.53 (95% CI: 0.46–0.68). The pooled DOR was 5.01 (95% CI:
response definitions provided by the eating disorder studies were
3.38–7.42). The combined AUC was 0.71 (S.E. = 0.03), which
similar in terms of the time cut-off, they varied with regard to the
suggests that it has a fair capacity to classifying later outcome.
symptom change required for early response classification. This
might explain the heterogeneity and inconsistency of accuracy
Discussion measures or even their discriminative function for the same
disorder.
The aim of this systematic review is to collate the literature
relating to the early response to treatment in eating disorders
and conduct a meta-analysis using diagnostic test accuracy Clinical implication
methodology to examine the robustness of the early response These findings suggest that for adolescent AN early weight gain
concept as a predictor of outcome. during the first four weeks of psychotherapeutic outpatient treat-
We were able to synthesize information from 24 studies. The ment is predictive of later improvements in eating disorder psy-
meta-analytic procedures suggest that the characterization of early chopathology and attainment of a healthy expected body weight.
response as a predictor of remission is more robust for AN and This holds true regardless of the psychotherapeutic protocol used.
BED, than for BN. Predicting a good response early in the course of treatment is
The high specificity and low sensitivity of the pooled accuracy particularly relevant for AN in which the ‘standard dose’ of
measures from AN studies suggest that a failure to respond to out- therapy can be high, for example, ranging from 15 × 1.5 h sessions
patient psychological treatment in the early phase is associated during 12 months (FBT, plus monitoring) in adolescents, through
with persistent symptoms at 1 year. The pooled AUC shows that to 40 sessions over 40 weeks in adults (CBT-E) and 20 weeks or
there is a 77% chance that a patient with an early response will more for inpatient care. If it is indeed possible to adapt treatment
have a remission of symptoms at the end of treatment using the in the early phase in order to produce change, this would be of
current criteria. The therapeutic input or ‘dose’ ranged across great therapeutic benefit. Previous research which investigated
studies, from three to ten sessions. Moreover, the time by which the effects of adding intensive parental coaching to standard
a specified reduction in symptoms was considered to be an ‘early FBT attested the beneficial impact of such an adaptive treatment
response’ varied across studies, ranging from the first week to the strategy (Lock et al., 2015). Other modules that might increase
first three months of treatment. These differences in the criteria the early response are adding a motivational intervention (Brewin
used for ‘early’ may explain some of the variability found in indi- et al., 2016) and modules to increase relatedness through includ-
vidual weight curves. ing families (Treasure & Nazar, 2016) and people who have
Furthermore, uncertainty surrounds how to define remission, recovered (Cardi et al., 2015). Additionally, adding modules to
and changing the parameters has been found to have a profound improving cognitive abilities (e.g. Cognitive Remediation
effect on outcome (Couturier & Lock, 2006). Most studies in Therapy; Tchanturia, Lounes, & Holttum, 2014) or training mod-
adolescent AN used a strict criteria (≥95% ideal body weight ules to increase social connection (Cardi et al., 2015) or eating be-
and scores within 1 standard deviation of Eating Disorders havior (Turton, Bruidegom, Cardi, Hirsch, & Treasure, 2016)
Examination community norms). might be of benefit.
In BN studies, early response did not strongly predict outcome. Also, the early development of a strong therapeutic aliance can
An early response in BN only increased the chance of predicting impact upon outcome in adolescentes receiving treatment for AN
remission by approximately 15%. Both sensitivity and specificity (Pereira, Lock, & Oggins, 2006), especially during the first four
were low. The definitions of early response ranged from a 51 to weeks of inpatient treatment (Sly, Morgan, Mountford, & Lacey,
95% reduction in binge/purge behaviours, whereas all studies 2013). Elsewhere, evidence suggests that interventions that focus
used abstinence from these behaviours as part of the definition on reducing shame and improving self-compassion early in

Eur. Eat. Disorders Rev. (2016) Copyright © 2016 John Wiley & Sons, Ltd and Eating Disorders Association.
Early Response to Eating Disorder Treatment B. P. Nazar et al.

treatment also aid in promoting a better outcome (Brewin et al., later outcome for AN and BED, but further work is needed to es-
2016; Kelly, Carter, & Borairi, 2014). tablish whether such an approach is reliable for BN.
For BED, results suggest that a reduction of at least 65% in
binge eating frequency at the fourth week of treatment (both psy- Acknowledgement
chotherapy alone or with adjunctive pharmacotherapy) predicts Janet Treasure is partly funded by the National Institute for
binge-eating abstinence at the end of treatment. Health Research (NIHR) Mental Health Biomedical Research
Centre at South London and Maudsley NHS Foundation Trust
Conclusion and King’s College London. The views expressed are those of
the author(s) and not necessarily those of the NHS, the NIHR
These studies show that current definitions of early behaviour or the Department of Health.
change have fair accuracy in predicting later symptomatic remis- Bruno Palazzo Nazar is supported by an international doctoral
sion in treatments used for AN and BED. There is less predictive scholarship from Coordenação de Aperfeiçoamento de Pessoal de
accuracy using the criteria presently used for BN treatment. It Nível Superior (CAPES), Brazil, and by the Federal University of
may be possible to adapt treatment in the early phase and improve Rio de Janeiro, Institute of Psychiatry (IPUB-UFRJ).

REFERENCES behavioral therapy. Behaviour Research and Therapy, 45(11), nervosa in a randomized clinical trial. The International Journal of
2537–2550. 10.1016/j.brat.2007.05.010. Eating Disorders, 39(8), 639–647. 10.1002/eat.20328.
Agras, W. S., Crow, S. J., Halmi, K. A., Mitchell, J. E., Wilson, G. T., Grilo, C. M., Masheb, R. M., & Wilson, G. T. (2006). Rapid response Lock, J., Le Grange, D., Agras, W. S., Fitzpatrick, K. K., Jo, B.,
& Kraemer, H. C. (2000). Outcome predictors for the cognitive to treatment for binge eating disorder. Journal of Consulting and Accurso, E., et al. (2015). Can adaptive treatment improve out-
behavior treatment of bulimia nervosa: Data from a multisite Clinical Psychology, 74(3), 602–613. 10.1037/0022-006X.74.3.602. comes in family-based therapy for adolescents with anorexia
study. The American Journal of Psychiatry, 157(8), 1302–1308. Grilo, C. M., White, M. A., Masheb, R. M., & Gueorguieva, R. nervosa? Feasibility and treatment effects of a multi-site treat-
10.1176/appi.ajp.157.8.1302. (2015). Predicting meaningful outcomes to medication and ment study. Behaviour Research and Therapy, 73, 90–95.
Agüera, Z., Riesco, N., Jiménez-Murcia, S., Islam, M. A., Granero, R., self-help treatments for binge-eating disorder in primary care: 10.1016/j.brat.2015.07.015.
Vicente, E., et al. (2013). Cognitive behaviour therapy response The significance of early rapid response. Journal of Consulting MacDonald, D. E., Trottier, K., McFarlane, T., & Olmsted, M. P.
and dropout rate across purging and nonpurging bulimia and Clinical Psychology, 83(2), 387–394. 10.1037/a0038635. (2015). Empirically defining rapid response to intensive treat-
nervosa and binge eating disorder: DSM-5 implications. BMC Grilo, C. M., White, M. A., Wilson, G. T., Gueorguieva, R., & ment to maximize prognostic utility for bulimia nervosa and
Psychiatry, 13(1), 285. 10.1186/1471-244X-13-285. Masheb, R. M. (2012). Rapid response predicts 12-month post- purging disorder. Behaviour Research and Therapy, 68, 48–53.
Attia, J. (2003). Moving beyond sensitivity and specificity: Using treatment outcomes in binge-eating disorder: Theoretical and 10.1016/j.brat.2015.03.007.
likelihood ratios to help interpret diagnostic tests. Australian clinical implications. Psychological Medicine, 42(4), 807–817. Madden, S., Miskovic-Wheatley, J., Wallis, A., Kohn, M., Hay, P., &
Prescriber, 26(5), 111–113. 10.1017/S0033291711001875. Touyz, S. (2015). Early weight gain in family-based treatment
Cardi, V., Ambwani, S., Crosby, R., Macdonald, P., Todd, G., Hartmann, A., Wirth, C., & Zeeck, A. (2007). Prediction of failure of predicts greater weight gain and remission at the end of treat-
Park, J., et al. (2015). Self-Help And Recovery guide for Eating inpatient treatment of anorexia nervosa from early weight gain. ment and remission at 12-month follow-up in adolescent an-
Disorders (SHARED): Study protocol for a randomized Psychotherapy Research, 17(2), 218–229. 10.1080/ orexia nervosa. The International Journal of Eating Disorders,
controlled trial. Trials, 16, 165. 10.1186/s13063-015-0701-6. 10503300600702315. 48(7), 919–922. 10.1002/eat.22414.
Cochrane handbook for DTA reviews. (2010). Retrieved from http:// Hilbert, A., Hildebrandt, T., Agras, W. S., Wilfley, D. E., & Wilson, Manea, L., Gilbody, S., & McMillan, D. (2015). A diagnostic
methods.cochrane.org/sdt/handbook-dta-reviews G. T. (2015). Rapid response in psychological treatments for meta-analysis of the Patient Health Questionnaire-9 (PHQ-9)
Couturier, J., & Lock, J. (2006). What is remission in adolescent binge eating disorder. Journal of Consulting and Clinical Psychol- algorithm scoring method as a screen for depression. General
anorexia nervosa? A review of various conceptualizations and ogy, 83(3), 649–654. 10.1037/ccp0000018. Hospital Psychiatry, 37(1), 67–75. 10.1016/j.genhosppsych.
quantitative analysis. International Journal of Eating Disorders, Kelly, A. C., Carter, J. C., & Borairi, S. (2014). Are improvements in 2014.09.009.
39(3), 175–183. 10.1002/eat.20224. shame and self-compassion early in eating disorders treatment Marrone, S., Mitchell, J. E., Crosby, R., Wonderlich, S., & Jollie-
Doyle, P. M., Le Grange, D., Loeb, K., Doyle, A. C., & Crosby, R. D. associated with better patient outcomes? The International Jour- Trottier, T. (2009). Predictors of response to cognitive behavioral
(2010). Early response to family-based treatment for adolescent nal of Eating Disorders, 47(1), 54–64. 10.1002/eat.22196. treatment for bulimia nervosa delivered via telemedicine versus
anorexia nervosa. The International Journal of Eating Disorders, Lalkhen, A. G., & McCluskey, A. (2008). Clinical tests: Sensitivity face-to-face. The International Journal of Eating Disorders, 42(3),
43(7), 659–662. 10.1002/eat.20764. and specificity: Fig 1. Continuing Education in Anaesthesia, Criti- 222–227. 10.1002/eat.20603.
Fairburn, C. G., Agras, W. S., Walsh, B. T., Wilson, G. T., & Stice, E. cal Care & Pain, 8(6), 221–223. 10.1093/bjaceaccp/mkn041. Masheb, R. M., & Grilo, C. M. (2007). Rapid response predicts treat-
(2004). Prediction of outcome in bulimia nervosa by early Le Grange, D. (2016). Anorexia nervosa in adults: The urgent need ment outcomes in binge eating disorder: Implications for stepped
change in treatment. The American Journal of Psychiatry, for novel outpatient treatments that work. Psychotherapy, Chi- care. Journal of Consulting and Clinical Psychology, 75(4),
161(12), 2322–2324. 10.1176/appi.ajp.161.12.2322. cago, Ill53(2), 251–254. 10.1037/pst0000057. 639–644. 10.1037/0022-006X.75.4.639.
Fernàndez-Aranda, F., Álvarez-Moya, E. M., Martínez-Viana, C., Le Grange, D., Accurso, E. C., Lock, J., Agras, S., & Bryson, S. W. Mitchell, A. J., Shukla, D., Ajumal, H. A., Stubbs, B., & Tahir, T. A.
Sànchez, I., Granero, R., Penelo, E., et al. (2009). Predictors (2014). Early weight gain predicts outcome in two treatments (2014). The Mini-Mental State Examination as a diagnostic and
of early change in bulimia nervosa after a brief for adolescent anorexia nervosa. The International Journal of Eat- screening test for delirium: Systematic review and meta-
psychoeducational therapy. Appetite, 52(3), 805–808. 10.1016/j. ing Disorders, 47(2), 124–129. 10.1002/eat.22221. analysis. General Hospital Psychiatry, 36(6), 627–633. 10.1016/j.
appet.2009.03.013. Le Grange, D., Doyle, P., Crosby, R. D., & Chen, E. (2008). Early genhosppsych.2014.09.003.
Fernández-Aranda, F., Krug, I., Jiménez-Murcia, S., Granero, R., response to treatment in adolescent bulimia nervosa. The Interna- Moher, D., Liberati, A., Tetzlaff, J., & Altman, D. G. (2009). Pre-
Núñez, A., Penelo, E., et al. (2009). Male eating disorders and tional Journal of Eating Disorders, 41(8), 755–757. 10.1002/eat.20566. ferred Reporting Items for Systematic Reviews and Meta-
therapy: A controlled pilot study with one year follow-up. Journal Linardon, J., Brennan, L., & de la Piedad Garcia, X. (2016). Rapid Analyses: The PRISMA statement. PLoS Medicine, 6(7)
of Behavior Therapy and Experimental Psychiatry, 40(3), 479–486. response to eating disorder treatment: A systematic review and e1000097. 10.1371/journal.pmed.1000097.
10.1016/j.jbtep.2009.06.004 meta-analysis. The International Journal of Eating Disorders, Pereira, T., Lock, J., & Oggins, J. (2006). Role of therapeutic alliance
Grilo, C. M., & Masheb, R. M. (2007). Rapid response predicts binge 49(10), 905–919. 10.1002/eat.22595. in family therapy for adolescent anorexia nervosa. The Interna-
eating and weight loss in binge eating disorder: Findings from a Lock, J., Couturier, J., Bryson, S., & Agras, S. (2006). Predictors of tional Journal of Eating Disorders, 39(8), 677–684. 10.1002/
controlled trial of orlistat with guided self-help cognitive dropout and remission in family therapy for adolescent anorexia eat.20303.

Eur. Eat. Disorders Rev. (2016) Copyright © 2016 John Wiley & Sons, Ltd and Eating Disorders Association.
B. P. Nazar et al. Early Response to Eating Disorder Treatment

Raykos, B. C., McEvoy, P. M., Erceg-Hurn, D., Byrne, S. M., Takwoingi, Y., Riley, R. D., & Deeks, J. J. (2015). Meta-analysis of Journal of the Eating Disorders Association, 22(1), 59–65.
Fursland, A., & Nathan, P. (2014). Therapeutic alliance in diagnostic accuracy studies in mental health. Evidence-Based 10.1002/erv.2262.
Enhanced Cognitive Behavioural Therapy for bulimia nervosa: Mental Health, 18(4), 103–109. 10.1136/eb-2015-102228. Wales, J., Brewin, N., Cashmore, R., Haycraft, E., Baggott, J., Cooper, A.,
Probably necessary but definitely insufficient. Behaviour Research Tchanturia, K., Lounes, N., & Holttum, S. (2014). Cognitive remedi- et al. (2016). Predictors of positive treatment outcome in people with
and Therapy, 57, 65–71. 10.1016/j.brat.2014.04.004. ation in anorexia nervosa and related conditions: A systematic anorexia nervosa treated in a specialized inpatient unit: The role of
Raykos, B. C., Watson, H. J., Fursland, A., Byrne, S. M., & Nathan, P. review. European Eating Disorders Review: The Journal of the early response to treatment. European Eating Disorders Review: The
(2013). Prognostic value of rapid response to enhanced cognitive Eating Disorders Association, 22(6), 454–462. 10.1002/erv.2326. Journal of the Eating Disorders Association. 10.1002/erv.2443.
behavioral therapy in a routine clinic sample of eating disorder Thompson-Brenner, H., Shingleton, R. M., Sauer-Zavala, S., Williams, S. E., Watts, T. K. O., & Wade, T. D. (2012). A review of
outpatients. International Journal of Eating Disorders, 46(8), Richards, L. K., & Pratt, E. M. (2015). Multiple measures of rapid the definitions of outcome used in the treatment of bulimia
764–770. 10.1002/eat.22169. response as predictors of remission in cognitive behavior therapy nervosa. Clinical Psychology Review, 32(4), 292–300. 10.1016/j.
Safer, D. L., & Joyce, E. E. (2011). Does rapid response to two group for bulimia nervosa. Behaviour Research and Therapy, 64, 9–14. cpr.2012.01.006.
psychotherapies for binge eating disorder predict abstinence? 10.1016/j.brat.2014.11.004. Wilson, G. T., Vitousek, K. M., & Loeb, K. L. (2000). Stepped care
Behaviour Research and Therapy, 49(5), 339–345. 10.1016/j. Treasure, J., & Russell, G. (2011). The case for early intervention in treatment for eating disorders. Journal of Consulting and Clinical
brat.2011.03.001. anorexia nervosa: Theoretical exploration of maintaining factors. Psychology, 68(4), 564–572.
Schlup, B., Meyer, A. H., & Munsch, S. (2010). A non-randomized The British Journal of Psychiatry: The Journal of Mental Science, Zamora, J., Abraira, V., Muriel, A., Khan, K., & Coomarasamy, A.
direct comparison of cognitive-behavioral short- and long-term 199(1), 5–7. 10.1192/bjp.bp.110.087585. (2006). Meta-DiSc: A software for meta-analysis of test accuracy
treatment for binge eating disorder. Obesity Facts, 3(4), Vall, E., & Wade, T. D. (2015). Predictors of treatment outcome in data. BMC Medical Research Methodology, 6(1), 31. 10.1186/
261–266. 10.1159/000319538. individuals with eating disorders: A systematic review and 1471-2288-6-31.
Sly, R., Morgan, J. F., Mountford, V. A., & Lacey, J. H. (2013). meta-analysis. The International Journal of Eating Disorders, Zunker, C., Peterson, C. B., Cao, L., Mitchell, J. E., Wonderlich, S.
Predicting premature termination of hospitalised treatment for 48(7), 946–971. 10.1002/eat.22411. A., Crow, S., et al. (2010). A receiver operator characteristics
anorexia nervosa: The roles of therapeutic alliance, motivation, Vaz, A. R., Conceição, E., & Machado, P. P. P. (2014). Early analysis of treatment outcome in binge eating disorder to identify
and behaviour change. Eating Behaviors, 14(2), 119–123. response as a predictor of success in guided self-help treatment patterns of rapid response. Behaviour Research and Therapy,
10.1016/j.eatbeh.2013.01.007. for bulimic disorders. European Eating Disorders Review: The 48(12), 1227–1231. 10.1016/j.brat.2010.08.007.

Supporting information

Additional Supporting Information may be found in the online version of this article at the publisher’s web site.

Eur. Eat. Disorders Rev. (2016) Copyright © 2016 John Wiley & Sons, Ltd and Eating Disorders Association.