REVIEW
Predictors of Treatment Outcome in Individuals with Eating Disorders: A Systematic Review and Meta-Analysis EvaVall, BA (Hons)* Tracey D. Wade, PhD
ABSTRACT Objective: Understanding the factors that predict a favourable outcome following specialist treatment for an eating disorder may assist in improving treatment efficacy, and in developing novel interventions. This review and meta-analysis examined predictors of treatment outcome and drop-out. Method: A literature search was conducted to identify research investigating predictors of outcome in individuals treated for an eating disorder. We organized predictors first by statistical type (simple, meditational, and moderational), and then by category. Average weighted mean effect sizes (r) were calculated for each category of predictor. Results: The most robust predictor of outcome at both end of treatment (EoT) and follow-up was the meditational mechanism of greater symptom change early during treatment. Simple baseline predictors associated with better outcomes at both EoT and follow-up included higher BMI, fewer binge/purge behaviors, greater motivation to recover, Resumen Objetivo: El entendimiento de los factores que predicen un resultado favorable despu es de un tratamiento especializado para un trastorno de la til conducta alimentaria, puede ser u para mejorar la eficacia de los tratamientos y para el desarrollo de intervenciones novedosas. Esta revisi on y meta an alisis examin o los factores que predicen los resultados de tratamiento y de abandono del mismo. squeda de M etodo: Se realiz o una bu la literatura existente en la investigaci on de los factores que predicen los resultados de tratamiento en individuos tratados por un trastorno de la conducta alimentaria. Organizamos dichos predictores, primero por tipo estadıstico (simple, media y moda) y
lower depression, lower shape/weight concern, fewer comorbidities, better interpersonal functioning and fewer familial problems. Drop-out was predicted by more binge/purge behaviors and lower motivation to recover. For most predictors, there was large interstudy variability in effect sizes, and outcomes were operationalized in different ways. There were generally insufficient studies to allow analysis of predictors by eating disorder subtype or treatment type. Discussion: To ensure that this area continues to develop with robust and clinically relevant findings, future studies should adopt a consistent definition of outcome and continue to examine complex multivariate predictor models. Growth in this area will allow for stronger conclusions to be drawn about the prediction of outcome for specific diagnoses and treatment types. Keywords: anorexia; bulimia; binge eating disorder; outcomes; response; predictor; moderator; mediator
luego por categorıas. El promedio pon~o del efecto (r), fueron derado y taman calculados para cada categorıa de factor de predicci on. Resultados: el m as fuerte factor de predicci on de los resultados tanto al Final del Tratamiento (EoT) como en el Seguimiento posterior, fue el mecanismo meditaci on de un mayor cambio de sıntomas tempranamente en el tratamiento. Los predictores simples basales asociados con mejores resultados tanto en EoT y seguimiento, incluyeron: IMC m as alto, disminuci on de las conductas de atrac on/purgaci on, mayor motivaci on para la recuperaci on, menor presencia de depresi on, menor preocupaci on acerca de la figura/peso, menos comorbilidades, mejor funciona-
Accepted 13 February 2015 *Correspondence to: Eva Vall, School of Psychology, Flinders University, PO Box 2100, Adelaide, South Australia 5001, Australia. E-mail:
[email protected] School of Psychology, Flinders University, Adelaide, South Australia, Australia Published online 14 July 2015 in Wiley Online Library (wileyonlinelibrary.com). DOI: 10.1002/eat.22411 C 2015 Wiley Periodicals, Inc. V
946
International Journal of Eating Disorders 48:7 946–971 2015
PREDICTORS OF TREATMENT OUTCOME IN INDIVIDUALS WITH EATING DISORDERS miento interpersonal y menos problemas familiares. Los predictores de Abandono del tratamiento fueron: m as conductas de atrac on/purgaci on y menor motivaci on al cambio. Para la mayorıa de los predictores, hubo una gran variabilidad inter-estudios ~o del efecto y los resultaen el taman dos cuando fueron operacionalizados en diferentes formas. En general, no hubo suficientes estudios para permi-
tir el an alisis de los predictores de acuerdo a subtipos de trastorno de la conducta alimentaria o tipo de tratamiento. n: Para asegurar que esta Discusio area contin ue desarroll andose con hallazgos clınicos robustos y relevantes, los estudios futuros deben adoptar una definici on coherente de resultados y continuar examinando los complejos y multivaria-
Introduction Identifying predictors of outcome is an important goal in the quest for improving outcomes in eating disorder treatment, where outcomes remain suboptimal even after intensive intervention.1 This is particularly true of outcomes related to receiving specialist treatment for an eating disorder (as opposed to naturalistic follow-up in the absence of an intervention) as, at the individual level this would allow us to identify those people who are most at risk of a poor outcome. This knowledge could be used to offera more targeted or intensive interventions to be offered to these individuals and to ensure that their long-term follow up is given priority. In the wider sense, knowing which factors affect treatment outcome is valuable because it offers specific insights into how treatments for eating disorders can be improved. Identification of factors that impede progress can inform the modification of existing treatments, or the development of future interventions. Despite the promise of this kind of approach, there is a well-documented paucity of robust and consistent findings related to predictors of response to treatment in eating disorders. While to our knowledge no systematic review of predictors of response to specialist treatment across eating disorders exists, several reviews of treatment outcomes more generally have considered this question. One review of anorexia nervosa (AN) treatment studies (n 5 35)2 attempted to identify sociodemographic predictors of treatment outcome, but found that evidence to support the predictive value of any factors linked to AN treatment outcome was weak. Several predictive factors for outcome for bulimia nervosa (BN) were identified in a review of randomised controlled trials (n 5 47) for BN treatment, including more frequent binge eating and longer illness duration.3 However, these effects were found for individuals receiving behavioral interventions and did not extend to medication trials. A more recent review of treatment studies (n 5 79) of BN treatment arrived at similar International Journal of Eating Disorders 48:7 946–971 2015
dos modelos predictivos. El crecimiento en esta area nos llevar a a conclusiones m as firmes acerca de la predicci on de resultados para diagn osticos y tipos de C tratamiento especıficos. V 2015 Wiley Periodicals, Inc (Int J Eat Disord 2015; 48:946–971).
conclusions, noting that although significant research efforts had been invested in this area, it had failed to identify consistent prognostic factors.4 Another review of studies (n 5 62) that considered outcomes for both AN and BN5 rated several factors as ‘moderately successful’ in predicting poorer treatment response, namely the presence of mood and anxiety disorders and impaired social functioning. It should be noted, however, that several studies included in the review were based on findings with participants who had been identified through community screening or assessment, and some of these people may not have received specialist treatment. Accordingly, for these studies it cannot be concluded that the predictors of outcome were associated with treatment per se, or whether they simply reflected the normal course of eating disorders. The results of a European collaboration involving over 2,000 patients with either AN, BN, or OSFED, treated across 12 countries in 80 treatment centres, suggested that greater symptom severity might be associated with a poorer outcome at 12 month followup.6 However, there are numerous challenges inherent across such a trial, and findings must therefore be interpreted with caution. Finally, a review of randomized controlled trials for BED (n 5 26) concluded that evidence to support specific predictors was “sparse.”7 Moreover, all the reviews noted that sample sizes varied widely across studies. Drop-out from treatment has also been studied as a specific type of outcome in the eating disorder field, and one review has specifically investigated predictors of drop-out.8 This review examined studies (n 5 26) reporting factors associated with dropout from treatment for all eating disorders and noted several predictors, including the binge/purge subtype of AN, styles of personality (i.e., low selfdirectedness and low cooperativeness), and psychological traits (i.e., high maturity fear and impulsivity). However, the authors noted that the ability to draw meaningful conclusions was hampered by methodological limitations across the studies, including small 947
VALL AND WADE
sample sizes and lack of replication of findings. A second review of studies (n 5 7) reporting drop-out from treatment for AN9 concluded that evidence to support the presence of robust predictors was both scarce and conflicting, but also found some evidence that individuals exhibiting more binge/purge behaviours were less likely to complete treatment. As in the reviews of treatment outcome, sample sizes varied widely across studies included in the reviews of treatment drop-out. Since the publication of the most recent reviews in 2009, a number of studies have emerged that have included at least some analysis of predictors of treatment response. The purpose of this review and meta-analysis is to systematically examine the existing literature across all eating disorders and present a rigorous summary of the evidence for predictors of treatment outcome in individuals with an eating disorder. In addition, it extends the insights offered by previous reviews in several ways. First, to our knowledge, no metaanalysis of predictors of treatment outcome or drop-out has been conducted, which limits the capacity to draw conclusions about the importance and clinical significance of potential predictors given different (i) numbers of studies examining any one predictor, and, (ii) varying sample sizes. Second, none of the reviews has investigated more complex predictor models, such as interactions between predictors, moderator or mediator effects. In this review, predictors are categorized first by the type of relationship they have with outcome (simple, mediational and moderational), and then by type. Third, it examines both predictors of drop-out from treatment and predictors of improvement after receiving specialist treatment for an eating disorder. Fourth, in line with the inclusion of binge eating disorder (BED) as an eating disorder in the latest edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM-5),10 this is the first review of predictors of treatment outcome that has included AN, BN and BED in one review, which is consistent with the trend to attempt to identify commonalities across diagnoses in order to inform transdiagnostic treatment approaches.11–13 Finally, given the large number of studies being conducted in the area, methodological issues that continue to limit the usefulness of findings in this area are examined and these considerations are consolidated to provide recommendations to improve the design of future treatment studies that also intend to examine predictors of outcome.
948
Method Information Source and Search Strategy The present study was conducted in accordance with the evidence-based guidelines for systematic reviews set forth in the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA).14 The primary search strategy involved searching peer reviewed published papers using a multi-field search in two databases, namely PsycINFO and PubMed. The final database search was conducted on 26 December 2014. The following combinations of search terms were employed where the terms appeared in either the title or abstract of the article: 1. Anorexia OR bulimia OR binge eating disorder OR eating disorder 2. Treatment 3. Response OR outcome and 4. Predictor OR predict OR mediator OR mediate OR moderator OR moderate The secondary search strategy involved identifying relevant articles from the reference lists from articles identified in the primary search. Inclusion Criteria The following specifications were applied: (i) Englishlanguage publication, (ii) eating disorder diagnosis, (iii) specialist eating disorder treatment, and (iv) published in a peer-reviewed journal. The search was limited to articles published in the last 30 years (i.e., since 1984). Study Selection Prior to examining the results, the search outputs from the two databases were first cross referenced and all duplicate records were removed. Next, the abstract of every record was reviewed to ensure that the inclusion criteria were met, and that the study related broadly to the review question. The full-text of all remaining records was examined to confirm eligibility in the qualitative synthesis. Finally, all studies were screened for inclusion in the meta-analysis, including the calculation of effect sizes. Studies were excluded if there was insufficient data to calculate an effect size. The authors discussed any studies where there was uncertainty about inclusion, and studies were only included if both authors agreed that they met inclusion criteria. A flow diagram of the selection process based on the PRISMA guidelines is presented in Figure 1. Data Extraction For each category of predictor, data was extracted from all studies addressing that variable, as shown in Table 1. This included the type of study (i.e., randomized controlled International Journal of Eating Disorders 48:7 946–971 2015
PREDICTORS OF TREATMENT OUTCOME IN INDIVIDUALS WITH EATING DISORDERS FIGURE 1.
PRISMA flow diagram of study selection.
trial or case series), eating disorder subtype, the number of individuals included in the analysis of the particular predictor, specific type of treatment(s) administered, an operationalized description of the outcome variable, the outcome period, and the specific predictor measure. Categorizing Predictors of Treatment Response For this study, predictors were categorized first in terms of the way in which they contribute to outcome, or their relationship with the outcome variable. To this end, predictors in the current review were first divided into three categories, namely simple predictors, mediators, and moderators. These were defined as follows: International Journal of Eating Disorders 48:7 946–971 2015
A simple predictor variable is one that is measured at baseline, and that directly predicts change in the outcome variable over time. A treatment mediator identifies possible mechanisms through which treatment impacts on outcome,15 and should be defined as a measure of an event or change that occurs after the onset of treatment.16 This can be either a change in the level of a baseline variable early in the course of treatment (for example, change in baseline depression during treatment could affect outcome) or change in another variable (for example, baseline depression could lead to increased anxiety during treatment, thus affecting outcome). Kraemer et al.15 explain
949
950
Design
Sample
n
Treatment
Characteristics of studies included in meta-analyses
Prepost RCT Prepost RCT Prepost
154 155 24 45 87
BED AN (adult & adolescent) AN, BN, OSFED BN AN AN BN 107 27 127 63 43
75 107
RCT Prepost
RCT Prepost
88 83
45 97
AN BN
BN BN
AN
63 37
101 48
484
CBT vs. SSCM CBT
CBT 1 2 forms of exposure Day hospital
Inpatient
Residential CBT vs. ERP Inpatient CBT vs. SSCM Outpatient CBT-E
CBTgsh vs. BWLgsh Inpatient or day hospital
EoT
30 Months
RCT
Prepost
Remission (no binge/purge in past 28 days) 157 RCT BN 120 CBT vs. SPT vs. Desipramine Remission (no binge/purge in last 2 weeks) Binge/purge behaviors predicting outcome at follow-up (fewer binge/purge behaviors 5 better outcome)
Naturalistic
87
95
BN/EDNOS-BN
BN
AN
186
43
63
42
Treatment as needed
Outpatient CBT-E
CBT vs. SSCM
Guided self help
Time to remission
QoL, mental health, depression
Remission
EoT
EoT
EoT
Reaching min 92.5% IBW by discharge EoT Reaching a BMI of 20 EoT
EoT
EoT
45
BN
Day hospital Inpatient
Binge vs.no binge groups
Overall ED pathology
Prepost
49 127
CBT vs.ERP
Sequential Tx vs. CBT
94
AN AN
27
110
Prepost Prepost
BN
BN
93 24
EoT EoT
EoT EoT
RCT
EDE-Q, binge frequency, depression BMI >= 18.5, no binge/purge in past 28 days, EDE global within two standard deviations of community norms Reduction in binges Binge/purge frequency
155
Inpatient or day hospital CBT or GSH (2 sites)
CBTgsh vs. BWLgsh Inpatient or day hospital
RCT
208 353
75 154
92
AN, BN BN
BED AN
12 months 2–12 months
12 months 2 years
2 years
EoT EoT EoT EoT EoT
EoT EoT
Prepost Prepost
RCT Prepost
QoL, mental health, depression Overall ED pathology
Recovery (weight restoration, nil excessive exercise, normal eating behavior, absence of excessive weight/shape concern) Binge/purge frequency Relapse in 2 years post-tx
EDE and EDI global scores Binge vs. no binge groups Reaching a BMI of 20 QoL, mental health, depression Remission (no binge/purge in past 28 days)
Binge frequency EDE-Q score
Time
111 91
85 89
Binge/purge behaviors predicting outcome at EoT (fewer binge/purge behaviors 5 better outcome)
Prepost
32
Eating disorder pathology predicting outcome at Follow-up
RCT RCT
85 86
Outcome Measure
Simple Predictors of Outcome
Eating disorder pathology predicting outcome at EoT (lower ED pathology 5 better outcome)
Reference
TABLE 1.
0.06 (20.19:0.30) 0.10 (20.23:0.41)
0.04 (20.16:0.24) 0.38 (0.11:0.60)
0.86 (0.82:0.90)
0.46 (0.30:0.60) 0.35 (20.04:0.64) 0.06 (20.12:0.23) 0.00 (20.25:0.25) 0.01 (20.29:0.31)
0.36 (0.15:0.54) 0.26 (20.08:0.43)
r (95% CI)
Pre-tx binge/purge frequency
Binges at intake Baseline binge/purge frequency Pretreatment binge frequency Pretreatment binge frequency AN subtype Having the binge/ purge AN subtype Pretreatment binge frequency Having the binge/ purge AN subtype Baseline purging frequency Weekly binge/purge frequency
0.18 (20.01:0.36)
0.02 (20.16:0.20)
0.23 (20.08:0.50)
0.10 (20.15:0.34)
0.05 (20.26:0.35)
20.08 (20.35:0.21) 0.16 (20.02:0.32)
0.33 (20.06:0.63)
0.20 (0.02:0.38)
0.38 (0.24:0.51) 0.36 (0.27:0.45)
Pre-tx binge frequency 0.19 (20.04:0.40) Having the binge/ 0.39 (0.25:0.52) purge AN subtype
EDI: bulimia subscale Eating attitudes test: bulimia scale EDE Global score Eating attitudes test
EDI total score
EDE-Q Global score Eating Attitudes Test EDI: 3 subscales EDE Global score EDI total score
EDE-Q Global score EDE-Q Global score
Predictor Measure
VALL AND WADE
International Journal of Eating Disorders 48:7 946–971 2015
International Journal of Eating Disorders 48:7 946–971 2015
Prepost
Prepost
Prepost
Prepost Prepost RCT
RCT
RCT
Prepost
Prepost
Naturalistic
RCT
RCT
Prepost
Prepost
158
96
27
109 110 45
92
159
83
97
95
52
25
89
94
Sample
110
105 127 63
BN
AN
AN
AN
AN/EDNOS-AN
BN
42
154
47
93
126
37
AN 158 (Adolescent) BN 48
BN
AN BN AN
AN 57 (Adolescent) AN 100
BN 80 (adolescent) AN, BN, OSFED 181
n
Prepost
RCT
26
45
BN (adolescent) AN
AN AN AN AN
63
40
113 83 50 49
Naturalistic RCT
Prepost
Prepost Prepost
95 52
96
109 160
126 93
AN 57 (Adolescent) AN 105 AN 700 (14 sites) (Adolescent)
AN/EDNOS-AN AN
BMI predicting outcome at follow-up (higher BMI 5 better outcome)
Prepost Prepost Prepost Prepost
111 38 112 93
BMI predicting outcome at EoT (higher BMI 5 better outcome)
RCT
Design
Continued
40
Reference
TABLE 1.
Inpatient Treatment as needed
Treatment as needed Inpatient then CBT 6 fluoxetine Inpatient
CBT vs. SSCM
Outpatient or day hospital
Inpatient or day hospital Inpatient Inpatient (completers) Day hospital
Guided self help
Inpatient or day hospital
Inpatient then CBT 6 fluoxetine Inpatient: TAU vs. TAU 1 MI
Treatment as needed
CBT
Day hospital
FBT vs. SyFT
Sequential Tx vs. CBT
Inpatient Internet self-help CBT vs. SSCM
Inpatient/Day hospital
Inpatient
Day hospital
FBT vs. SPT
Treatment
Change in BMI Weight recovery to 90% MBMI
Recovered vs. nonrecovered
Time to remission Weight maintenance
QoL, mental health, depression
Decrease in weekly binges, EDI scales
Weight gain Reaching a BMI of 17.5 Discharge BMI Reaching min 92.5% IBW by discharge
Readmission within 3 months of discharge Remission
Change in EDE score
Weight maintenance
Time to remission
Overall ED pathology
Relapse in 2 years post-tx
Achieving 95% of IBW
Overall ED pathology
Change in EDE global Change in EDI-2 Bulimia subscale QoL, mental health, depression
Relapse in 1 year post-tx
Recovered vs. nonrecovered
Rapid response vs. no response
Partial remission
Outcome Measure
Simple Predictors of Outcome Time
Predictor Measure Pre-tx binge/purge frequency Pre-tx binges
6 months 12 months
12 months
30 months 6–12 months
EoT
EoT
EoT EoT EoT EoT
Baseline weight Pre-Tx BMI
Pre-Tx BMI
Pre-Tx BMI Baseline BMI
Pre-Tx BMI
Intake BMI Pre-Tx BMI Pre-Tx BMI Being 6 85% IBW at admission Baseline BMI
Baseline purging behavior 12 months Having the binge/ purge AN subtype 6 months Fewer baseline SBEs 2 months EDI-2 bulimia subscale 6–12 Months Having the binge/ purge AN subtype 18 Months Pretreatment binge frequency 12 Months Having the binge/ purge AN subtype 2 years Pre-tx vomiting frequency 12 months Pre-tx vomiting frequency 30 months Pre-tx binge/purge frequency 6 2 12 months Having the binge/ purge AN subtype 6 weeks Pretreatment binge/ purge frequency 3 months Having the binge/ purge AN subtype 6 months Pretreatment binge frequency
4th session during Tx 12 months
6 months
0.49 (0.33: 0.62) 0.14 (0.07:0.21)
0.46 (0.23:0.64)
0.11 (20.15:0.36) 0.46 (0.28:0.61)
0.24 (20.01:0.46)
0.12 (20.20:0.42)
0.63 (0.50:0.73) 0.28 (0.07:0.47) 0.67 (0.48:0.80) 0.07 (20.22:0.34)
0.02 (20.29:0.32)
0.07 (20.09:0.23)
0.08 (20.21:0.36)
0.07 (20.14:0.27)
0.02 (20.24:0.28)
0.41 (0.10:0.64)
0.41 (0.14:0.62)
0.16 (0.00:0.31)
0.26 (0.08:0.43)
0.24 (0.05:0.41) 0.18 (0.01:0.34) 0.48 (0.26:0.65)
0.28 (0.09:0.45)
0.45 (0.22:0.64)
0.27 (0.10:0.43)
0.01 (20.21:0.23)
r (95% CI)
PREDICTORS OF TREATMENT OUTCOME IN INDIVIDUALS WITH EATING DISORDERS
951
952
Prepost Naturalistic Prepost Prepost RCT Prepost
Design
Continued
BN, BED BN/EDNOS-BN AN AN AN BN
Sample 42 186 79 41 63 37
n
Prepost
Prepost
Prepost
Prepost RCT
Prepost
Prepost
RCT
114
115
26
30 82
86
26
45
BN (adolescent) AN
AN (adolescent) AN (adolescent) BN (adolescent) BN AN (adult and adolescent) AN, BN, OSFED
AN
63
40
107
87 107
40
65
42
127
CBT vs. SSCM
Outpatient or day hospital
Residential
Guided selfhelp Inpatient or day hospital
Outpatient or day hospital
Inpatient
Outpatient CBT
Inpatient
Prepost
RCT RCT Prepost
161
25 45 161
47 155
AN 47 AN 63 AN, BN, OSFED 97
AN, BN, OSFED 97
AN BN spectrum disorders
Inpatient: TAU vs. TAU 1 MI CBT vs. SSCM Inpatient or day hospital
Inpatient or day hospital
Inpatient: TAU vs. TAU 1 MI Outpatient, day program, day hospital
RCT
Prepost
RCT
40
26
85
BN (adolescent) BN (adolescent) BED 75
40
80
CBTgsh vs. BWLgsh
Outpatient or day hospital
FBT vs. SPT
Depression predicting outcome at EoT (lower depression 5 better outcome)
RCT Prepost
25 116
Motivation predicting outcome at follow-up (greater motivation 5 better outcome)
Prepost
24
Increase in EDI interoceptive awareness EDE-Q, binge frequency, depression
Partial remission
Change in EDE score QoL, mental health, depression EDE-Q global score
EDE-Q global score
Change in EDE score Weekly binge/purge, ED pathology, mental health
QoL, mental health, depression
Time
EoT
EoT EoT
EoT
EoT
During Tx
EoT
6 months 30 months 12 months 12 months 6–12 months 2–12 months
ANSOCQ
Readiness to change Confidence in ability to change YBC-EDS motivation scale BNSOCQ
BNSOCQ
1 question re change readiness 6 motivational questions ANSOCQ
Lowest BMI Admission BMI Admission BMI Admission weight Pre-Tx BMI Pre-Tx BMI
Predictor Measure
EoT
EoT
EoT
0.14 (20.15:0.41) 0.06 (20.19:0.30) 0.04 (20.16:0.24)
0.18 (20.02:0.37)
0.27 (20.02:0.52) 0.28 (0.06:0.47)
0.08 (20.17:0.32)
0.20 (20.12:0.48)
0.25 (0.06:0.42)
0.25 (0.04:0.44) 0.30 (0.11:0.46)
0.45 (0.11:0.69)
0.39 (0.15:0.58)
0.17 (216:0.47)
0.22 (0.05:0.38)
0.35 (0.05 2 0.59) 0.01 (20.22:0.24) 0.05 (20.17:0.26) 0.19 (20.12:0.47) 0.18 (20.07:0.41) 0.24 (20.09:0.52)
r (95% CI)
Beck Depression 0.02 (20.20:0.24) Inventory Beck Depression 0.58 (0.33:0.76) Inventory MDD (SCID-I) and Beck 0.33 (0.11:0.52) Depression Inventory
All likert scales Autonomous and Controlled Motivations for Treatment Questionnaire 3 months Autonomous motivation (ACMT Questionnaire) 6 weeks ANSOCQ 6 2 12 months ANSOCQ 3 months Controlled motivation (ACMT questionnaire)
6 weeks 16 weeks
EoT
Increase in EDI interoceptive subscale EoT
EDE and EDI global scores
Overevaluation of weight/shape EDE-Q score
Overall improvement (incl. weight gain, QoL) Decrease in weekly binges/EDI scales
Weight change at 6 weeks
Reaching a BMI of 20
Full remission Time to remission Clinical deterioration (CGI-S score) % IBW QoL, mental health, depression Overall ED pathology
Outcome Measure
Simple Predictors of Outcome
Guided self help Treatment as needed Inpatient Inpatient CBT vs. SSCM CBT
Treatment
Motivation predicting outcome at EoT (greater motivation 5 better outcome)
94 95 113 66 45 97
Reference
TABLE 1.
VALL AND WADE
International Journal of Eating Disorders 48:7 946–971 2015
Prepost
Prepost
Prepost
Prepost
RCT
87
117
26
94
45
AN
BN (adolescent) BN
AN
BN
BN
Sample
63
42
40
63
43
27
n
CBT vs. SSCM
Guided self help
International Journal of Eating Disorders 48:7 946–971 2015
Prepost Prepost
Prepost
Prepost
RCT
97 117
94
118
45
AN
AN, BN
BN
BN AN
BN
63
102
42
37 63
101
CBT vs. SSCM
Not described
Guided self help
CBT Inpatient CBT-E
CBT 1 2 forms of exposure
RCT
Prepost
162
91
BN
BN
BED
353
24
75
CBT or GSH (2 sites)
CBT vs. focal psychotherapy
CBTgsh vs. BWLgsh
RCT
RCT
120
162
BN
BN 24
75
AN, BN, OSFED 57
CBT vs. focal psychotherapy
CBT vs. behavioral CBT vs. IPT
Outpatient CBT-E
121
Prepost
AN
103
Outpatient CBT
Shape/weight concern predicting outcome at EoT (lower shape/weight concern 5 better outcome)
Prepost
119
Self-esteem predicting outcome at follow-up (greater self-esteem 5 better outcome)
RCT
85
Self-esteem predicting outcome at EoT (greater self-esteem 5 better outcome)
RCT
88
Recovery: no longer meeting DSM-IV criteria
General psychopathology (PSE)
Montgomery and Asberg depression rating scale Beck depression inventory Brief symptom inventory depression subscale Beck depression inventory Beck depression inventor Beck depression inventor
Predictor Measure
EoT
EoT
EoT
0.04 (20.21:0.29)
0.15 (20.05:0.34)
0.03 (20.28:0.33)
0.26 (20.07:0.54) 0.17 (20.11:0.42)
0.33 (0.14:0.49)
0.08 (20.17:0.32)
0.05 (20.26:0.35)
0.25 (20.07:0.52)
0.18 (20.08:0.43)
0.25 (20.05:0.51)
0.36 (20.02:0.65)
r (95% CI)
EoT
EDE-Q shape concern
0.27 (0.09:0.44)
Rosenberg self esteem 0.31 (0.04:0.54) scale 8 2 12 months Rosenberg self esteem 0.66 (0.35:0.84) scale
Rosenberg self esteem 0.10 (20.16:0.35) scale
Rosenberg self esteem 0.35 (0.13:0.54) scale Rosenberg self esteem 0.57 (0.22:0.79) scale Rosenberg self esteem 0.05 (20.06:0.15) scale
Presence of major depression (Hamilton depression rating scale) 12 months Illness duration 6 2 12 months Brief symptom inventory depression subscale 6 months Beck depression inventor 3 years Beck depression inventory 6 2 12 months Beck depression inventor
12 months
EoT
EoT
EoT
EoT
EoT
EoT
Time
Positive (no longer meeting DSM-IV 4 weeks criteria) vs. negative outcome (meeting DSM-IV criteria or premature drop-out) EDE scores 12 months
Purge frequency
General psychopathology (PSE)
EDE-Q, binge frequency, depression
QoL, mental health
Psychiatric status rating scale (PSRS)
Remission
Overall ED pathology BMI, EDE scores
Binge/purge frequency
QoL, mental health
Remission
Decrease in weekly binges/EDI scales
Remission (no binge/purge in past 28 days) BMI, EDE scores
Binge vs. no binge groups
Outcome Measure
Simple Predictors of Outcome
Outpatient or day hospital
Inpatient CBT-E
Outpatient CBT-E
CBT vs. ERP
Treatment
Depression predicting outcome at follow-up (lower depression 5 better outcome)
RCT
Design
Continued
155
Reference
TABLE 1.
PREDICTORS OF TREATMENT OUTCOME IN INDIVIDUALS WITH EATING DISORDERS
953
954
Prepost RCT
Prepost
37 122
91
BN
AN, BN BED
BED
Sample
353
50 116
108
n CBT vs. Fluoxetine vs. CBT 1 Fluoxetine Inpatient Group CBT 1 Individual CBT and/or Fluoxetine CBT or GSH (2 sites)
Treatment
Binge/purge frequency
EDI change Binge eating severity
Remission from bingeing
Outcome Measure
Simple Predictors of Outcome
Naturalistic
RCT
95
120
BN
AN/EDNOS-AN
BN/EDNOS-BN
75
126
186 Time to remission
Time to remission
CBT vs.behavioural CBT vs. IPT EDE score and bulimic behaviors
Treatment as needed
Treatment as needed
EoT
EoT
CBT vs. SSCM
Prepost Prepost (retrospective) RCT
37 124
85
Prepost
123
AN, BN 50 AN 268 (Adolescent) BED 75
AN, BN, OSFED 93
CBTgsh vs. BWLgsh
Inpatient Inpatient
Day program
Comorbid psychopathology predicting outcome at EoT (lower comorbidity 5 better outcome)
63
Inpatient
EDE-Q, binge frequency, depression
Change in EDI interoceptive awareness Change in EDI over treatment Length of stay
QoL, mental health, depression
Partial or full recovery
EoT
EoT EoT
EoT
6–12 months
RCT
95
45
AN (adolescent) AN
3 months
EoT
EoT
10–15 years
QoL, mental health, depression
EDE-Q score
Reduction in binges/binge abstinence EoT
Abstinence from binge eating
Achieving BMI of at least 20 at EoT and no binge/purge during last 4 weeks of treatment
Prepost
CBT vs. SSCM
Inpatient or day hospital
Inpatient or day hospital
Group CBT vs. Group IPT
Inpatient/day hospital
98
63
107
152
162
218
12 months
AN (adult and adolescent) AN
AN-BP, BN
BED
AN
12 months
30 months
30 months
EoT
EoT EOT
EoT
Time
Interpersonal functioning predicting outcome at follow-up (better interpersonal functioning 5 better outcome) 90 RCT BN 43 Inpatient vs. day hospital Failure vs. success (remission or partial remission) 34 RCT BED 162 Group CBT vs. Group IPT Abstinence from binge eating
RCT
45
Prepost
111
RCT
RCT
34
86
Prepost
27
Interpersonal functioning predicting outcome at EoT (better interpersonal functioning 5 better outcome)
Naturalistic
95
Shape/weight concern predicting outcome at follow-up (lower shape/weight concern 5 better outcome)
RCT
Design
Reference
43
Continued
TABLE 1.
SCL-90R Not specified: based on patient charts Any Axis I comorbidity (SCID-I)
SCL-90R
Social Adjustment Scale Inventory of Interpersonal Problems Premorbid social relating Weissman social adjustment scale
Inventory of Interpersonal Problems: Social Inhibition subscale Inventory of interpersonal problems Inventory of interpersonal problems: social avoidance subscale Social quality of life score Weissman social adjustment scale
EDE shape/weight concern EDE shape/weight concern EDE weight and shape concern subscales
EDE weight and shape concern subscales
EDE weight and shape concern subscales Body attitudes test Two items from EDE
Predictor Measure
0.09 (20.14:0.31)
0.30 (0.02:0.53) 0.27 (0.16:0.38)
0.29 (0.09:0.47)
0.09 (20.16:0.33)
0.33 (0.14:0.50)
0.25 (0.09:0.40)
0.37 (0.07:0.60)
0.27 (0.02:0.49)
0.18 (20.01:0.36)
0.24 (0.08:0.39)
0.25 (0.09:0.40)
0.16 (0.03:0.29)
0.36 (0.09:0.58)
0.05 (20.21:0.30)
0.15 (20.04:0.33)
0.18 (0.08:0.28)
0.46 (0.21:0.66) 0.26 (0.08:0.42)
0.28 (0.09:0.44)
r (95% CI)
VALL AND WADE
International Journal of Eating Disorders 48:7 946–971 2015
Design
Continued
Sample
n
Treatment
Weight maintenance
Time to remission
Naturalistic
RCT
RCT
Prepost RCT RCT
95
125
126
127 99 128
71
International Journal of Eating Disorders 48:7 946–971 2015 CBT vs. Desipramine vs. CBT 1 Desipramine Inpatient Group CBT vs. Group IPT CBT 6 exposure
CBT 6 exposure
Treatment as needed
Prepost
RCT Prepost
RCT Prepost
RCT
64
45 124
92 26
85
AN (Adolescent) AN AN (Adolescent) BN BN (adolescent) BED
BED
75
110 40
63 268
65
108
CBTgsh vs. BWLgsh
Sequential Tx vs. CBT Outpatient or day hospital
CBT vs. SSCM Inpatient
CBT vs.Fluoxetine vs. CBT 1 Fluoxetine Outpatient FBT
EDE-Q, binge frequency, depression
Overall ED pathology Weekly binges, EDI scores
Achieving 95% of IBW at end of treatment QoL, mental health, depression Length of stay
Remission from bingeing
Change in EDE scores Binge days & global EDE Presence of ED diagnosis
Abstinence from purging
Any ED diagnosis
Prepost
RCT RCT
Prepost Prepost (from 2 studies) Prepost
RCT
Prepost
129
92 159
97 100
101
52
96
93
44
37 177
110 158
32
AN 57 (Adolescent)
AN
BN
BN AN (Adolescent) BN BN
BED
Inpatient then CBT 6 fluoxetine Inpatient
CBT
CBT CBT 6 medication
Sequential Tx vs. CBT FBT vs. SyFT
DBT
Recovered vs. nonrecovered
Recovered vs. nonrecovered (MAEDS scale) Weight maintenance
Overall ED pathology Remission for 6 months
Abstinence from bingeing in past month Overall ED pathology Achieving 95% of IBW
Age of onset/illness duration predicting outcome at follow-up (shorter illness duration 5 better outcome)
RCT
43
Age of onset/illness duration predicting outcome at EoT (shorter illness duration 5 better outcome)
AN, BN, OSFED 77 BED 162 BN 134
BN
AN, BN, OSFED 312 (adult and adolescent) BN 109
Change in EDI-2 Bulimia subscale Time to remission
Inpatient then CBT 6 fluoxetine Personality Disorders predicting outcome at follow-up (fewer PD symptoms 5 better outcome)
93
Internet self-help Treatment as needed
RCT
127 312
52
Recovered vs. nonrecovered
Prepost Naturalistic
Inpatient
110 95
57
Prepost
96
AN (Adolescent) BN AN, BN, OSFED (adult and adolescent) AN
Outcome Measure
Simple Predictors of Outcome
Comorbid psychopathology predicting outcome at follow-up (lower comorbidity 5 better outcome)
Reference
TABLE 1.
12 months
0.61 (0.49:0.71) 0.02 (20.09:0.13)
0.31 (0.05:0.53)
r (95% CI)
Illness duration Illness duration and age of onset Illness duration
Illness duration Illness duration
Early age of onset
Age of onset
Illness duration Duration of disorder
Duration of AN Duration of AN
Illness duration
Age of onset
Avoidant PD (SCID-II) Any PD (SCID-II) Any PD (SCID-II)
Total PD symptoms (SCID-II) Cluster B score (PDE)
Any PD (SCID-II)
Duration of AN and age of onset
0.23 (20.03:0.46)
0.03 (20.18:0.23)
0.08 (20.22:0.37)
0.36 (0.04:0.61) 0.17 (0.02:0.31)
0.22 (0.03:0.39) 0.16 (0.00:0.31)
0.21 (9:35.32:0.38)
0.09 (9:35.14:0.31)
0.30 (0.12:0.46) 0.13 (9:35.19:0.42)
0.35 (0.11:0.55) 0.17 (0.05:0.28)
0.32 (0.08:0.53)
0.01 (9:35.18:0.20)
0.14 (20.09:35) 0.16 (0.01:0.31) 0.08 (9:35.09:0.25)
0.37 (0.14:0.56)
0.01 (20.18:0.20)
0.13 (0.02:0.24)
Number of Axis I diag- 0.02 (20.19:0.22) noses (SCID-I)
SCL-90R Any Axis I comorbidity (SCID-I)
CIDI and SCL-90
Predictor Measure
6 2 12 months AN duration
9 years
12 months 11.5 years
18 months 12 months
6 months
EoT
EoT EoT
EoT EoT
EoT
EoT
5 years 12 months 3 years
4–12 months
5 years
30 months
6–12 months
2 months 30 months
12 months
Time
PREDICTORS OF TREATMENT OUTCOME IN INDIVIDUALS WITH EATING DISORDERS
955
956
RCT
45
AN
AN
Sample
63
47
n
Prepost RCT
Prepost RCT
RCT
23 131
132 133
133
BED
BN BN
AN BN
BN
RCT
Prepost
Prepost
135
109
98
AN (adolescent) AN
AN
105
121
53
259
117 128
185 105
182
Inpatient
FBT vs.AFT
Inpatient CBT-E
Outpatient CBT-E Outpatient CBT (therapist-led, therapist-assisted, or selfhelp versions) Outpatient CBT (face-to-face or telemedicine)
Inpatient/day hospital Outpatient CBT 6 exposure
Outpatient CBT-E
CBT vs. SSCM
RCT
Prepost
Prepost
137
138
139
AN, BN (adolescent) BN
AN (adolescent) BN
51
49
94
62
Outpatient CBT
Treatment as needed
Imipramine vs. group CBT
2 versions of FBT
RCT Prepost
RCT
RCT
137 138
65
159
68
94 49
AN 158 (Adolescent)
BN AN, BN (adolescent) AN (adolescent) FBT vs. SyFT
Long vs. short term family therapy
Imipramine vs. group CBT Treatment as needed
Familial factors predicting outcome at follow-up (fewer familial problems 5 better outcome)
RCT
136
Achieving 95% of IBW
Remission (95% IBW 1 EDE within two SD of community norms
Remission Average outcome score
Predictor Measure
EoT
EoT
EoT
10 2 15 years
6 months
EoT
EoT
EoT
EoT EoT
12 months
12 months
10 years 12 months
Maternal depression Maternal critical comments Lower family expressiveness and cohesion Not having an intact family
Paternal substance abuse Maternal critical comments Controlling family environment (Family environment scale)
Parental warmth
Less intense exercise at baseline Compulsive drive to exercise at EoT
Physical activity duration/expenditure Driven exercise
Weight suppression
Weight suppression Weight suppression
Last 4 weeks of Weight suppression Tx EoT Weight suppression EoT Weight suppression
Duration of AN and age of onset 6 2 12 months Duration of AN
Time 6 weeks
Change in EDI, bingeing and vomiting EoT
Average outcome score
Remission
Weight at or above 95% IBW
Partial or full recovery
Change in BMI
EDE global score
Not completing treatment program
Reduction in bingeing, abstinence from bingeing
Weight gain and bulimic symptoms Weight gain over Tx, binge/purge frequency Binge/purge frequency Reduction in bingeing, abstinence from bingeing
Binge eating frequency
QoL, mental health, depression
Change in EDE score
Outcome Measure
Simple Predictors of Outcome
Inpatient: TAU vs. TAU 1 MI
Treatment
AN 95 Inpatient (adolescent) Familial factors predicting outcome at EoT (fewer familial problems 5 better outcome)
Prepost
134
Exercise (less exercise 5 better outcome)
Prepost
130
Weight suppression (lower weight suppression 5 better outcome)
RCT
Design
Reference
25
Continued
TABLE 1.
0.19 (0.04:0.34)
0.27 (0.03:0.50)
0.56 (0.40:0.69) 0.38 (0.11:0.60)
0.44 (0.19:0.64)
0.53 (0.29:0.71)
0.25 (0.05:0.43)
0.29 (0.04:0.50)
0.37 (0.18:0.53)
0.49 (0.33: 0.62)
0.34 (0.17:0.49)
0.42 (0.17:0.62)
0.02(20.10:0.14)
0.01 (20.17:0.19) 0.01 (20.16:0.18)
0.13 (20.02:0.27) 0.06 (20.13:0.25)
0.14 (20.01:0.28)
0.05 (20.20:0.29)
0.06 (20.23:0.34)
r (95% CI)
VALL AND WADE
International Journal of Eating Disorders 48:7 946–971 2015
Design
Reference
Sample
n
International Journal of Eating Disorders 48:7 946–971 2015
Prepost
RCT
Prepost
RCT
RCT
102
131
132
133
133
BED
BN
BN
BN
BN
BN
259
128
117
132
32
182
Prepost
Prepost
141
142
Prepost
Prepost
RCT
Prepost
Prepost
Prepost
Prepost
Prepost
143
61
144
103
104
134
24
89
188
AN
AN
AN
AN
AN
AN
AN
AN
154
127
53
213
166
63
218
81
AN, BN, OSFED 65
AN, BN, OSFED 125
BN
AN subtype (binge/purge 5 more drop-out)
Prepost
140
Binge/purge behaviours (more binge/purge 5 more drop-out)
Prepost
130
Weight suppression (higher weight suppression 5 more drop-out)
Predictors of drop-out
Continued
TABLE 1.
Inpatient or day hospital
Inpatient
Inpatient CBT-E
Inpatient
Inpatient
CBT-AN vs. SSCM
Inpatient/day hospital
Inpatient
Day hospital
Day hospital
Outpatient CBT (multisite)
Outpatient CBT (therapist-led, therapist-assisted, or selfhelp versions) Outpatient CBT (face-to-face or telemedicine)
Outpatient CBT-E
Outpatient CBT 6 exposure
Outpatient CBT-E
Outpatient CBT-E
Treatment
n/a
n/a
Discharge before reaching a BMI of 20 Discharge against medical advice
Not completing treatment program
Discharge against medical advice
Discharge prior to reaching target weight of 90% of IBW and maintaining it for a minimum of 2 weeks Not achieving BMI of at least 20 at EoT and continued binge/purge during last 4 weeks of treatment Discharge before completing at least 30 sessions Discharge prior to reaching BMI > 20
Not completing full course of treatment Not completing full 4 months of treatment Discharge prior to reaching goal weight and demonstrating ability to maintain it
Not completing Tx
n/a
n/a
n/a
n/a
n/a
n/a
n/a
n/a
Having the binge/ purge AN subtype Having the binge/ purge AN subtype Having the binge/ purge AN subtype Having the binge/ purge AN subtype Having the binge/ purge AN subtype Having the binge/ purge AN subtype
Having the binge/ purge AN subtype
Having the binge/ purge AN subtype
History of purging
n/a
Highest adult weight pre Tx weight
Highest adult weight pre Tx weight Highest adult weight pre Tx weight
Highest previous weight - pre-Tx weight Highest ever weight lowest ever weight Highest adult weight pre Tx weight
Predictor Measure
n/a
Time
Bulimic thoughts questionnaire EDI bulimia subscale
n/a
n/a
Completion 8 CBT sessions and mini- n/a mum 5 of 8 behavior therapy sessions Nonmutual premature termination of n/a treatment Not completing Tx n/a
Not completing Tx
Not completing Tx
Outcome Measure
Simple Predictors of Outcome
0.14 (20.02:0.29)
0.16 (20.02:0.32)
0.00 (20.27:0.27)
0.24 (0.11:0.36)
0.38 (0.24:0.51)
0.36 (0.12:0.56)
0.23 (0.00: 0.46)
0.07 (20.15:0.28)
0.25 (0.00:0.46)
0.26 (20.41: 20.08)
0.29 (0.15:0.42)
0.02 (20.10:0.14)
0.01 (20.16:0.18)
0.03 (20.15:0.21)
0.03 (20.14:0.20)
0.37 (0.03:0.64)
0.61 (0.49:0.70)
r (95% CI)
PREDICTORS OF TREATMENT OUTCOME IN INDIVIDUALS WITH EATING DISORDERS
957
958
AN
Sample
Prepost
RCT
Prepost Prepost
65
106 107
RCT
Prepost
Prepost
85
108
102
75
BN
AN
BED
32
133
75
AN 86 (adolescent) AN, BN, OSFED 186 BN 111
AN
Depression (greater depression 5 more drop-out)
Prepost
148
Comorbid psychopathology (higher comorbidity 5 more drop-out)
AN, BN, OSFED 261
236
Outpatient CBT
Inpatient
CBTgsh vs.BWLgsh
Long vs.hort term family therapy Inpatient Outpatient CBT
Inpatient CBT
Outpatient & day hospital
Outpatient CBT 1 nutritional rehabilitation
105
AN
188
Prepost
BN
145
Guided self-help
Outpatient CBT (multisite)
40
Inpatient
Outpatient CBT
Prepost
Prepost
146
65
42
Inpatient
140
Prepost
115
AN (adolescent) AN (adolescent) BN
127
Outpatient CBT 1 nutritional rehabilitation
Treatment
CBT vs. IPT
Prepost
114
AN
236
n
Not completing Tx
Termination prior to planned discharge date
Not explicity described
Being administratively discharged Termination prior to session 10
Completing 80% of assigned sessions
Any discontinuation of Tx
Any unplanned interruption of therapy before reaching target weight (BMI > 19) Not completing Tx
Not completing full course of treatment
Not completing full Tx program
Not completing Tx program
Length of stay
Discharge before reaching a BMI of 20 Not remaining in Tx at 6 weeks
Any unplanned interruption of therapy before reaching target weight (BMI > 19)
Outcome Measure
Simple Predictors of Outcome
147 RCT BN 129 Impulsivity (greater impulsivity 5 more drop-out)
Prepost
24
Motivation (higher motivation 5 less drop-out)
Prepost
Design
Reference
145
Continued
TABLE 1.
n/a
n/a
n/a
n/a n/a
n/a
n/a
n/a
n/a
n/a
n/a
n/a
n/a
n/a
n/a
n/a
Time
0.29 (0.05:0.50)
0.40 (0.11:0.63)
0.24 (0.07:0.40)
0.12 (20.01:0.24)
r (95% CI)
Negative affect (Beck Depression Inventory) Not having a current major depressive episode Higher pre-tx depression
Schizophrenic, hypomanic or deviant/ antisocial features (MMPI) Comorbid psychopathology (K-SADS) SCID-I comorbidity Having had previous psychiatric treatment
Higher impulsivity (EDI impusle regulation)
0.37 (0.03:0.64)
20.23 (20.39:0.06)
0.31 (0.09:0.50)
0.19 (0.05:0.33) 0.21 (0.03:0.38)
0.16 (20.05:0.36)
0.02 (20.21:0.25)
0.14 (0.02:0.26)
Impulsivity scale of 0.26 (0.12:0.39) multidimensional personality questionnaire Self injurious behavior 0.20 (0.08:0.32)
Motivational Inventory 0.19 (20.13:0.47) measure Stage of change scale 0.14 (20.03:0.31)
1 motivational question 6 motivational questions ANSOCQ
Having the binge/ purge AN subtype
Predictor Measure
VALL AND WADE
International Journal of Eating Disorders 48:7 946–971 2015
Design
Reference
Sample
n
Treatment
International Journal of Eating Disorders 48:7 946–971 2015
Prepost
Prepost
RCT
RCT
RCT
RCT
Prepost
Prepost
Prepost
Prepost
Prepost
94
64
45
45
149
150
47
151
112
87
87
179
116
53
50
65
42
140
BN
BN
AN
BN
43
43
50
112
AN, BN, OSFED 105
BED
AN (Adolescent) AN (Adolescent) AN (Adolescent) BN
BN, BED
BN
Outpatient CBT-E
Outpatient CBT-E
Inpatient (drop-outs)
Outpatient CBT
Outpatient CBT
Telemedicine CBT vs.Face to face CBT CBT: therapist assisted vs. led vs. self-help
AFT
FBT
Outpatient FBT
Guided self help
Outpatient CBT (multisite)
Prepost
Prepost
RCT
RCT
RCT
94
113
152
149
153
BED
BN
BN
AN
BN, BED
101
58
220
79
42
DBT vs. active comparison
Telemedicine CBT
CBT vs. IPT
Inpatient
Guided self help
No binge/purge in past 28 days
Cessation of binge/purge
EDE score
Clinical deterioration (CGI-S score)
Full remission
Remission (no binge/purge in past 28 days) Remission (no binge/purge in past 28 days)
Discharge BMI
EoT
Remission (BMI 18.5 kg/m2, global EDE-Q within 1 SD of community norms, abstinence from binge/ purge for 28 days) EDE-Q global score
ED pathology at midtx (session 10) Weekly weight gain first half Tx 65% reduction in purging at week 4 % Reduction in EDI score by week 4
Change in purging after 4 weeks (six sessions) 51% change in binge frequencies at session 3 Min. 2.88% weight gain by session 4 Weight gain at session 8 Weight gain at session 5 Reduction in binge eating at week 8 15% reduction in binge eating by week 1 Response at approx week 4.6
Predictor Measure
51% change in binge frequencies at session 3 12 months Rate of weight gain (>0.8 kg/week) 8 months Percentage reduction in purging in first 4 weeks of Tx 3 2 12 months Reduction in binge eating at week 4 12 months 65% or greater reduction in OBE days by session 4
Time
6 months
EoT
EoT
EoT
EoT
EoT
EoT
EoT
EoT
EoT
EoT
EoT
Abstinence from bingeing at end of treatment
Achieving 95% of IBW at end of treatment Achieving 95% EBW and within 1 SD of community norms on EDE Achieving 95% EBW and within 1 SD of community norms on EDE Cessation of binge/purge
Full remission
Cessation of binge/purge
Outcome Measure
Simple Predictors of Outcome
Early symptom change predicting outcome at follow-up (greater change 5 better outcome)
Prepost
140
Early symptom change predicting outcome at EoT (greater change 5 better outcome)
Mediators of outcome
Continued
TABLE 1.
0.30 (0.11:0.47)
0.31 (0.06:0.53)
0.38 (0.26:0.49)
0.24 (0.02:0.43)
0.66 (0.45:0.80)
0.28 (20.02:0.54)
0.39 (0.10:0.62)
0.47 (0.22:0.66)
0.63 (0.50:0.73)
0.37 (0.18:0.53)
0.35 (0.21:0.47)
0.45 (0.29:0.58)
0.50 (0.27:0.68)
0.72 (0.55:0.83)
0.30 (0.06:0.51)
0.86 (0.75:0.92)
0.55 (0.41:0.67)
r (95% CI)
PREDICTORS OF TREATMENT OUTCOME IN INDIVIDUALS WITH EATING DISORDERS
959
0.04 (20.18:0.26)
0.08 (20.14:20.29)
0.37 (0.07:0.61) 0.17 (20.20: 20.50)
0.48 (0.27:0.65) 0.52 (0.36:0.65)
12 months 12 months
12 months Higher BMI and lower ED pathology
12 months
4.6 years 3 years
BMI > 18 and resumption of menses Improved EDE-Q weight and shape concern % IBW at follow-up BMI > 19
Less clinical deterioration (CGI-S score)
12 months Remission (95% IBW 1 EDE within two SD of community norms
Discharge BMI Increase in BMI from baseline to EoT % IBW at discharge Weight gain to >85.8% of IBW Achieving 95.2% of expected IBW Discharge BMI
0.29 (0.05:0.49)
0.24 (20.02:0.47)
Early adherence to meal plan Change in weight at week 9 2 years Time to relapse
Outcome Measure
Following, this grouping by relationship with outcome, predictors that were examined in at least three studies were grouped by category (e.g., eating disorder behaviors, personality disorders). Predictors that appeared in only one or two studies were grouped in a residual category termed “miscellaneous predictors.” The full table including predictors not included in the meta-analysis is available on request from the first author.
Prepost
AN 83 (Adolescent) AN 79 RCT
Inpatient
Inpatient CBT vs. medication vs. combination FBT vs. AFT AN AN Prepost RCT
41 28
Inpatient Outpatient CBT 66 103 AN AN Prepost Prepost
In some cases, multiple effect sizes from the same study were calculated. This was to indicate where a single predictor was used to predict more than one type of outcome (e.g., BMI and quality of life), or where an outcome was measured at more than one time point (e.g., at end of treatment and at 12 month follow-up). It is not recommended to include multiple effects from a single study in the meta-analytic process, as this increases the risk of a single study biasing the results.17 In order to mitigate the influence of such biases, all effects were first divided into those measuring outcome at the end of treatment (EoT), and those measuring outcome at a post-treatment followup. Next, for studies where multiple effects were present for a single predictor or outcome, these effects were combined into a single effect size. For example, if separate effect sizes were given for distinct outcomes measures (e.g., BMI and ED pathology), these were amalgamated. Similarly, if a study measured an outcome at both 6- and 12-month follow-up, these two effects were combined into a single follow-up effect size. The final study groupings used in the analyses is given in Table 1.
113
156
66 156
Statistical Analysis 51 121
AN 68 Long vs.hort term family (adolescent) therapy Discharge BMI predicting outcome at follow-up (higher BMI 5 better outcome) RCT 65
Day hospital 46
Prepost
AN, BN, OSFED 58
n Sample Design Reference
Continued TABLE 1.
960
that operationally, one would document temporal precedence (with the independent variable preceding the mediator), correlation between these two variables, and when one considered the two variables jointly, either total mediation or partial mediation. Treatment moderators identify for whom and under what circumstances treatments have effects,15 that is, variables that influence the strength or the direction of a relationship between a predictor variable and an outcome variable. By definition, a moderator is measured at pre-treatment and has no correlation with the treatment condition at baseline. Apart from an interaction between the moderator and predictor variable predicting outcome, a main effect between the moderator and the outcome may also exist. Moderators are often studied in terms of their relationship with treatment type, that is, to determine whether a baseline characteristic makes a particular type of treatment more or less beneficial.
Grouping Effects for the Meta-Analysis
Treatment
Simple Predictors of Outcome
Time
Predictor Measure
r (95% CI)
VALL AND WADE
Effect Sizes. For each outcome variable, an effect size expressed as the correlation coefficient, r, was calculated. International Journal of Eating Disorders 48:7 946–971 2015
PREDICTORS OF TREATMENT OUTCOME IN INDIVIDUALS WITH EATING DISORDERS TABLE 2.
Results of meta-analysis for each predictor variable k
Simple predictors of drop-out Higher weight suppression 6 Higher binge/purge frequency 3 Having binge/purge AN subtype 9 Lower motivation 5 Higher impulsivity 3 Greater comorbid psychopathology 4 Greater depressive symptoms 3 Simple predictors of better outcome at EoT Lower ED pathology 7 Lower binge/purge frequency 12 Higher BMI 6 Greater motivation to recover 9 Lower depression 9 Higher self-esteem 3 Lower shape/weight concern 5 Better interpersonal functioning 5 Lower comorbid psychopathology 4 Shorter duration/lower onset age 7 Lower weight suppression 6 4 Less exercisea Fewer familial problems 4 Simple predictors of better outcome at follow-up Lower ED pathology 5 Lower binge/purge frequency 17 Higher BMI 11 Greater motivation to recover 6 Lower depression 6 Higher self-esteem 3 Lower shape/weight concern 3 Better interpersonal functioning 4 Lower comorbid psychopathology 4 Absence of personality disorders 6 Shorter duration/lower onset age 10 Fewer familial problems 4 Mediators of better outcome at EoT Early symptom change 12 Mediators of better outcome at follow-up Early symptom change 7 Discharge BMI 6
Mean r
Variance
95% CI
Z
p
Q
p
Failsafe N
0.19 0.27 0.20 0.23 0.19 0.16 0.18
0.094 0.007 0.006 0.005 0.009 0.004 0.148
20.07:0.42 0.18:0.37 0.13:0.27 0.13:0.32 0.12:0.27 0.07:0.25 20.28:0.56
1.46 5.39 5.21 4.58 5.11 3.46 0.75
0.145