Initial searches
The initial search (conducted 29 January 2024) identified 7160 studies. A further three studies were identified from other sources (Fig. 1). After removal of duplicates, 4771 studies remained. Two reviewers (JP and CJ) screened 20.3% of titles and abstracts and one reviewer (JP) screened the remaining titles and abstracts, excluding 4607 studies and leaving 164 studies remaining. Two reviewers (JP and MR) then screened 22.6% full texts and at least one reviewer (JP or MR) screened the remaining full texts, excluding a further 135 studies. Supplementary Information S1.6. provides a summary of excluded studies with reasons. Key reasons for exclusion were insufficient data, no acute augmentation, and augmentation was a long-term combination treatment. Any discrepancies were resolved by discussion between JP, CJ, MR, and senior author TW. A final 29 studies met inclusion
criteria and were included in the systematic review [19,26–53]. Of these, 17 randomised trials that had eating disorder outcomes were included in the meta-analysis [19,26–41].
Meta-analysis
The final sample for the meta-analysis consisted of 1179 individuals with a mean age of 28.92 years (SD = 8.39; female = 95.9%) across 17 studies (Table 1). Research was most frequently conducted in the United States (n = 4; 23.5%) and the United Kingdom (n = 4; 23.5%), but with a broad spread including Australia (n = 2; 11.8%), Canada (n = 2; 11.8%), the Netherlands (n = 2; 11.8%), Italy (n = 1; 5.9%), New Zealand (n = 1; 5.9%), and Spain (n = 1; 5.9%). A summary of acute augmentations included in the meta-analysis appears in Supplementary Information S2.1.; each study has a different specific type of acute augmentation.
A meta-analysis of randomised controlled trials was conducted for eating disorder outcomes (n = 17), most frequently the global score from the Eating Disorder Examination Questionnaire. Results in Table 2 show a significant effect of acute augmentations to psychological therapy on eating disorder outcomes, with a small effect size. Q and I2 statistics revealed no heterogeneity beyond chance between studies. Egger’s test revealed no publication bias for eating disorder outcomes. A forest plot for pooled effects (pre-post differences across augmentation-control groups) for each study appears in Supplementary Information S2.2.
Weight status (i.e., underweight versus non-underweight based on majority eating disorder diagnosis) was tested as a potential moderator, in subgroup analyses. The meta-analysis was repeated with studies coded as underweight (n = 6) and non-underweight (n = 11). Results appear in Table 2, showing a significant effect of acute augmentations on eating disorder outcomes for both underweight and non-underweight samples, with small effect
Table 1
Characteristics of studies included in the meta-analysis (n = 17) and studies included in the systematic review only (n = 12).
Authors (year of publication) (country) | Included N (1) Augmentation & (2) Control group | Mean (SD) age (1) Augmentation & (2) Control group | Gender (% female) | ED Dx (%) | Weight Status | Psychological ED Measure | Interventions (1) Augmentation & (2) Control group | Pre-Post Ax Period |
Studies included in the meta-analysis (n = 17) |
Agras et al. (1989) [26]a (United States) | (1) 16 (2) 17 | 29.2 (8.6)b | 100 | BN = 100 | Non-Underweight | EAT DIET + EDI DTd | (1) CBT + response prevention (2) CBT | 4 months |
Anastasiadou et al. (2020) [27] (Spain) | (1) 53 (2) 53 | (1) 17.25 (3.54) (2) 18.88 (7.77) | 91.4 | AN-R = 48.1, AN-P = 6.6; BN = 12.3, BED = 3.8, OSFED = 29.2 | Underweight | EDE-Q global | (1) TAU + TCApp (2) TAU | 12 weeks |
Boerhout et al. (2017) [28] (Netherlands) | (1) 38 (2) 32 | (1) 25.00 (7.59) (2) 23.34 (7.30) | 94.3 | AN = 55.7; BN = 32.9; BED = 4.3; OSFED = 7.1 | Underweight | EDE-Q global | (1) TAU + body and movement-oriented intervention (2) TAU | 6 weeks |
Bulik et al. (1998) [19]a, f (New Zealand) | (1) 37 (2) 39 | 26.10 (6.10)b | NR | BN = 100 | Non-underweight | EDI DT | (1) CBT + B-ERP (2) Relaxation training | 14 weeks |
(1) 35 (2) 39 | 26.10 (6.10)b | NR | BN = 100 | Non-underweight | EDI DT | (1) CBT + P-ERP (2) Relaxation training | 14 weeks |
Cardi et al. (2019) [29] (United Kingdom) | (1) 99 (2) 88 | (1) 26.60 (8.46) (2) 29.15 (10.03) | 96.8 | AN or AAN (breakdown NR) | Underweight | EDE-Q global | (1) TAU + RecoveryMANTRA (2) TAU | 6 weeks |
Cesa et al. (2013) [30] (Italy) | (1) 27 (2) 20 | (1) 32.90 (8.80) (2) 29.9 (7.95) | 100 | BED = 100 | Non-underweight | BIAQ | (1) CBT + VR-enhanced CBT (2) CBT | 6 weeks |
Dingemans et al. (2013) [31] (Netherlands) | (1) 41 (2) 41 | (1) 27.00 (8.60) (2) 26.90 (8.00) | 100 | AN-R = 46.3; AN-BP = 25.6; BN = 11.0; EDNOS AN-R = 9.8; EDNOS AN-BP = 7.3 | Underweight | EDE-Q global | (1) TAU + CRT (2) TAU | 6 weeks |
Flynn et al. (2023) [32]a (United Kingdom) | (1) 20 (2) 20 | (1) 40.40 (9.83) (2) 43.70 (10.60) | 97.6 | BED = 100 | Non-underweight | EDE-Q global | (1) ABM Training + real tDCS (2) ABM Training | NR (2–3 weeks of Tx) |
Goldstein et al. (2014) [33] (Australia) | (1) 28 (2) 29 | (1) 23.43 (7.51) (2) 22.97 (6.72) | 98.4 | AN-R = 24.6; AN-BP = 15.8; BN = 24.6; EDNOS-R = 10.5; EDNOS-BP = 24.6 | Non-underweight | EDE-Q global | (1) TAU + CBT-P (2) TAU | NR (3.5 weeks of Tx) |
Grilo et al. (2013) [34] (United States) | (1) 24 (2) 24 | (1) 45.00 (11.80) (2) 46.50 (10.20) | 79.2 | BED = 100 | Non-underweight | EDE-Q global | (1) Usual care + SH-CBT (2) Usual care | 4 months |
Keeler et al. (2022) [35] (United Kingdom) | (1) 40 (2) 40 | (1) 30.00 (23.00, 40.75)c (2) 29.00 (23.00, 35.00)c | 96.3 | BN = 66.0; BED = 34.0 | Non-underweight | EDE-Q global | (1) TAU + FoodT app (2) TAU | 4 weeks |
Keshen et al. (2020) [36] (Canada) | (1) 40 (2) 41 | (1) 27.90 (9.11) (2) 27.64 (10.39) | 100 | AN 18.5; AN-BP 21.0%; BN 38.3; OSFED 8.6; UFED 13.6 | Non-underweight | EDE-Q global | (1) TAU + Recovery Record (2) TAU | NR (up to 32 weeks of Tx) |
Presseller et al. (2022) [37] (United States) | (1) 29 (2) 26 | (1) 38.07 (13.59) (2) 39.58 (15.04) | 83.6 | BN spectrum = 100 | Non-underweight | EDE global | (1) CBT-E + JITAIs (2) CBT-E | 16 weeks |
Reyes-Rodriguez et al. (2021) [38] (United States) | (1) 12 (2) 13 | (1) 39.80 (8.40) (2) 35.40 (9.40) | 100 | BN-NP = 28.0; BN-P = 20.0; BED = 28.0; EDNOS = 24.0 | Non-underweight | EDE-Q global | (1) CBT + family enhanced model (2) CBT | NR (28 sessions of Tx) |
Rowlands et al. (2022) [39] (United Kingdom) | (1) 37 (2) 30 | (1) 16.36 (1.59) (2) 15.47 (1.78) | 98.0 | AN = 92.0; BN = 4.0; EDNOS = 2.0; ARFID = 2.0 | Underweight | EDE-Q global | (1) TAU + online CBMT (2) TAU | Range = 3–24 weeks |
Trottier et al. (2015) [40] (Canada) | (1) 22 (2) 21 | 24.8 (7.3)b | 100 | AN = 26.7; BN = 64.4; EDNOS = 8.9 | Non-underweight | EDE WC | (1) Maintenance TAU + graded body image exposure (2) Maintenance TAU | NR (4–16 weeks of Tx) |
Wade et al. (2009) [41] (Australia) | (1) 22 (2) 25 | (1) 24.28 (6.37) (2) 19.71 (3.10) | 95.7 | AN = 100 | Underweight | EDE global | (1) TAU + MI (2) TAU | 2 weeks |
Studies included in the systematic review only (n = 12) |
Allen et al. (2012) [42] (Australia) | (1) 52 (2) 43 | (1) 26.52 (8.98) (2) 26.44 (8.98) | NR | AN = 22.1; BN = 32.6; EDNOS “AN-like” = 35.8; EDNOS “BN-like” = 6.3; EDNOS-P = 2.1; UFED = 1.1 | Non-underweight | — | (1) TAU + MFT (2) TAU | NR (range of Tx = 10–100 sessions) |
Children's Hospital of Philadelphia (2019) [43]a, f, g (United States) | (1) 19 (2) 21 | (1) 15.25 (1.61) (2) 15.51 (1.81) | 62.1 | AN = 100 | Underweight | — | (1) FBT + adolescent-focused CRT (2) FBT | 6 months |
(1) 19 (2) 21 | (1) 15.21 (1.53) (2) 15.51 (1.81) | 62.1 | AN = 100 | Underweight | — | (1) FBT + parent-focused CRT (2) FBT | 6 months |
Compare et al. (2016) [44] (Italy) | (1) 63 (2) 55 | (1) 51.10 (4.10) (2) 50.80 (6.00) | 57.6 | BED = 100 | Non-underweight | — | (1) EFGT + dietary counselling (2) EFGT | 20 weeks |
Fernandez-Aranda et al. (2015) [45] (Spain) | (1) 20 (2) 18 | 29.50 (9.90)b | 100 | BN = 100 | Non-underweight | — | (1) TAU + SVG (2) TAU | 16 weeks |
Galasso et al. (2020) [46] (Italy) | (1) 10 (2) 9 | (1) 54.00 (11.00) (2) 53.00 (13.00) | 100 | BN = 100 | Non-underweight | — | (1) TAU + CAAET (2) TAU | 6 months |
Ganci et al. (2018) [47] (Australia) | (1) 40 (2) 40 | (1) 15.20 (1.70) (2) 15.20 (1.50) | 80.0 | AN or ARFID (breakdown NR) | Underweight | — | (1) FBT + parent education and skills workshop (2) FBT | NR (> 12 weeks) |
Golan (2022) [48] (Israel) | (1) 47 (2) 47 | (1) 24.40 (0.86) (2) 24.20 (3.10) | 100 | AN = 37.0; BN = 41.0; BED = 11.0; ARFID = < 10.0 | Non-underweight | — | (1) MBT + ECOSA (2) MBT | 8 months |
Lackner et al. (2016) [49] (Austria) | (1) 10 (2) 12 | NR (range = 12–18 years) | 100 | AN = 100 | Underweight | — | (1) TAU + EEG neurofeedback training (2) TAU | 5 weeks |
Nakazato et al. (2009) [50] (Japan) | (1) 16 (2) 25 | (1) 23.2 (4.2) (2) 23.8 (5.9) | 97.6 | AN-BP = 14.6; BN-P = 63.4; BN-NP = 19.5; EDNOS = 2.4 | Non-underweight | — | (1) CBT + MET (2) CBT | 15 weeks |
Pendleton et al. (2002) [51]a (United States) | (1) 20 (2) 17 | 45.00 (8.30)b | 100 | BED = 100 | Non-underweight | — | (1) CBT + exercise intervention (2) CBT | 4 months |
Rigaud et al. (2011) [52] (France) | (1) 52 (2) 51 | (1) 27.4 (8.1) (2) 27.9 (6.2) | 100 | AN = 35.0; BN = 65.0 | Non-underweight | — | (1) CBT + tube feeding (2) CBT | 2 months |
Wilson et al. (1991) [53] | 22e | (1) 19.80 (SD NR) (2) 21.60 (SD NR) | NR | BN = 100 | Non-underweight | — | (1) CBT + ERP (2) CBT | 20 weeks |
Notes. N = sample size; SD = standard deviation; ED Dx = eating disorder diagnosis; ED = eating disorder; Pre-Post Ax period = pre-treatment to post-treatment assessment period; NR = not reported; Tx = treatment. |
a study includes additional groups not reported in the present study.
b indicates the mean age (SD) for the sample.
c indicates the median age (interquartile range) for the sample.
d indicates the average of the EAT Dieting subscale and the EDI Drive for Thinness subscale.
e indicates N for the sample.
f study included two independent acute augmentation groups compared to control; individuals were only counted once in descriptives.
g author represents clinical trial sponsor as reported on ClinicalTrials.gov as does not report author(s).
ED Dx: AAN = atypical anorexia nervosa; AN = anorexia nervosa; AN-BP = anorexia nervosa, binge-purge type; AN-P = anorexia nervosa, purging type; AN-R = anorexia nervosa, restricting type; ARFID = avoidant/restrictive food intake disorder; BED = binge eating disorder; BN = bulimia nervosa; BN-NP = bulimia nervosa, non-purging type; BN-P = bulimia nervosa, purging type; EDNOS = eating disorder not otherwise specified; EDNOS “AN-like” = eating disorder not otherwise specified, anorexia nervosa; EDNOS AN-BP = eating disorder not otherwise specified, anorexia nervosa, binge-purge type; EDNOS AN-R = eating disorder not otherwise specified, anorexia nervosa, restricting type; EDNOS “BN-like” = eating disorder not otherwise specified, bulimia nervosa; EDNOS-BP = eating disorder not otherwise specified, binge-purge type; EDNOS-P = eating disorder not otherwise specified, purging disorder; EDNOS-R = eating disorder not otherwise specified, restricting type; OSFED = other specified feeding or eating disorder; UFED = unspecified feeding or eating disorder. Categories were labelled as described by each study.
Psychological ED measures: BIAQ = Body Image Avoidance Questionnaire; EAT DIET = Eating Attitudes Test, Dieting subscale; EDE global = Eating Disorder Examination, global score; EDE WC = Eating Disorder Examination, Weight Concern subscale; EDE-Q global = Eating Disorder Examination-Questionnaire, global score; EDI DT = Eating Disorder Inventory, Drive for Thinness subscale.
Psychological therapies, augmentation, and control groups: ABM = Attention Bias Modification Training; CAAET = Combined Aerobic and Anaerobic Exercise Training; CBMT = Cognitive Bias Modification Training for Social Stimuli; CBT = Cognitive Behavioural Therapy; CBT-E = Enhanced Cognitive Behavioural Therapy; CBT-P = Cognitive Behavioural Therapy for Perfectionism; SH-CBT = Self-Help Cognitive Behavioural Therapy; CRT = Cognitive Remediation Therapy; ECOSA = Eating and Control Styles Axis; EEG = Electroencephalogram; EFGT = Emotionally Focused Group Therapy; ERP = Exposure with Response Prevention; B-ERP = Exposure with Response Prevention to Pre-Binge Cues; P-ERP = Exposure with Response Prevention to Pre-Purge Cues; JITAIs = Just-In-Time Adaptive Interventions; Maintenance TAU = maintenance treatment as usual; MET = Motivational Enhancement Therapy; MFT = Motivation-Focused Therapy; MI = Motivational Interviewing; RecoveryMANTRA = Guided Self-Help Intervention from Maudsley Model of Anorexia Nervosa Treatments for Adults; SVG = Serious Video Games; TAU = treatment as usual; tDCS = Transcranial Direct Current Stimulation; VR-enhanced CBT = Virtual Reality-Enhanced Cognitive Behavioural Therapy. TAU could consist of any of the following: CBT-based treatment (e.g., individual or group-based CBT or CBT-E), individual or group psychotherapy, family psychotherapy, outpatient or community care (e.g., visits with primary care physician or other healthcare providers, after care at home, medication management, symptom monitoring), inpatient care, intensive treatment program, day or day program treatment, nutrition or dietary counselling, behavior-oriented weight gain programme, or input from a multidisciplinary team (e.g., psychiatrist, psychologist, psychomotor therapist, nurse practitioner, nurse, dietician, nutritionist). Usual care could consist of care from primary care physician or other healthcare providers.
Table 2
Meta-analysis results for all studies and by weight status.
| k (N studies) | Effect size (95% CI) | p | I2 between studies (95% CI) | Cochran’s Q | Eggers’ test |
All Studies | 17 | .16 (.04, .29) | .02 | 0.0% (0.0%., 51.1%) | 15.06, p = .52 | p = .35 |
Subgroup analysis by weight status (underweight vs non-underweight)a |
| k (N studies) | Effect size (95% CI) | p | I2 between studies | Cochran’s Q | Eggers’ test |
Underweight samples | 6 | .15 (.07, .23) | < .01 | 0.0% | 0.70 | N/A |
Non-underweight samples | 11 | .18 (-.06, .42) | < .01 | 30.0% | 14.29 | N/A |
Notes. Effect size = Hedges’ g; CI = confidence interval. |
a 95% CI for I2 and p for Cochran’s Q not available.
sizes. However, the non-underweight sample had 95% confidence intervals for the effect size that crossed zero. Q and I2 statistics revealed no heterogeneity between underweight samples but mild heterogeneity for non-underweight samples.
Systematic review
The final sample for the systematic review consisted of 1907 individuals with a mean age of 29.44 years (SD = 10.55; female = 93.5%) across 29 studies (Table 1). Research was most frequently conducted in the United States (n = 6; 20.7%) but with wide representation including Australia (n = 4; 13.8%), the United Kingdom (n = 4; 13.8%), Italy (n = 3; 10.3%), Canada (n = 2; 6.9%), the Netherlands (n = 2; 6.9%), Spain (n = 2; 6.9%), Austria (n = 1; 3.4%), France (n = 1; 3.4%), Israel (n = 1; 3.4%), Japan (n = 1; 3.4%) and New Zealand (n = 1; 3.4%). Eight of the 12 studies not included in the meta-analysis contained participants with just one eating disorder diagnosis. A summary of acute augmentations included in the systematic review appears in Supplementary Information S2.3. Of note, across these studies, subgroups of augmentation could be identified (i.e., augmentations used in more than one study). The most investigated was forms of motivational enhancement, followed by App-based recording of dietary intake, movement-related interventions, cognitive remediation therapy, family-based interventions, and exposure and response prevention.