Evidence Based Update on Psychosocial Treatments for Eating Disorders in Children and Adolescents

ABSTRACT

Eating disorders (EDs) are life-threatening psychiatric illnesses that occur in adolescents. Unfortunately, limited randomized controlled trials exist to address EDs in this vulnerable population. The current review updates a prior Journal of Clinical Child and Adolescent Psychology review from 2015. The recommendations in this review build upon those that were previously published. This update was completed through a systematic search of three major scientific databases (PsychInfo, Pubmed, and Cochrane) from 2015 to 2022 (inclusively) from three databases, employing relevant medial subject headings. Additionally, expert colleagues were asked for additional literature to include. Thirty-one new studies were added to this review. Psychosocial treatments included family therapies, individual therapy, cognitive-behavioral therapy, interpersonal psychotherapy, cognitive training, dialectical behavioral therapy, and more recently, virtual or telehealth-based practices and guided self-help modalities for carers of youth with EDs. Using the Journal of Clinical Child and Adolescent Psychology’s methodological review criteria, this update found behavioral family-based treatment modalities (FBT) for both adolescent anorexia nervosa and bulimia nervosa met well-established treatment criteria. To date, there were no well-established treatments found for child and adolescent avoidant-restrictive food intake disorder, or binge eating disorder. Internet facilitated cognitive-behavioral therapy and family-based therapy were found to be possibly efficacious for binge eating disorder. Family-based treatment was found to be possibly efficacious for avoidant restrictive food intake disorder, with other clinical trials for cognitive treatment modalities under way. Ongoing research examining treatments for eating disorders in children and adolescents broadly is needed.

Introduction

Eating disorders are life-threatening psychological illnesses that onset most frequently during adolescence. The lifetime prevalence of anorexia nervosa (AN) is approximately .48–1.7% in adolescent females (Hoek & van Hoeken, 2003; Lucas et al., 1991; Pinhas et al., 2011; Smink et al., 2014; Van Son et al., 2006), and the prevalence of bulimia nervosa (BN) is approximately 1–2% in females (Ahs et al., 2006; Fairburn & Beglin, 1990). The data reporting prevalence rates of AN in males is limited, but BN rates are approximately .5% in male youth (Carlat et al., 1997). The estimated female-to-male ratio of ED diagnoses is 10:1 (Lock, 2009). Notably, there are a significant number of subthreshold cases of eating disorders amongst adolescents, demonstrating ED behaviors, and psychopathology comparable to those with DSM-5 diagnoses. Rates of binge eating disorder in youth are approximately 2.3% in females and 2.6% in males (Swanson et al., 2011). The most recently incorporated ED diagnosis, Avoidant Restrictive Food Intake Disorder (ARFID), more frequently onsets in early childhood and has higher prevalence rates of approximately 7.2–17.4%. Sex-specific prevalence rates have not yet been determined given the relative newness of the diagnosis (Wons Thomas et al., 2018). Altogether, comorbid diagnoses are frequent in adolescents with EDs, with 50% of adolescents reporting a comorbid affective disorder and 35% reporting a comorbid anxiety disorder (Godart et al., 2002; Holtkamp et al., 2005).

The health consequences of eating disorders can be severe and result in medical stabilization hospitalization. The aggregate mortality rate per decade in EDs has been found to be an alarming 5.6% (Arcelus et al., 2011; Franko et al., 2013). The medical instability seen within the context of an eating disorder involves the physiological impacts of behaviors that maintain EDs (including, but not limited to, restrictive or selective eating, over-exercise, and/or purging) and can result in bradycardia, hypotension, and orthostatic hypotension (Modan-Moses et al., 2003; Olmos et al., 2010). In AN, cardiac arrest accounts for mortality in 50% of cases of EDs, while suicidality accounts for the other 50% (Birmingham et al., 2005). Mortality rates for the other ED categories are less studied, although some research suggests BN and AN have similar mortality rates (Crow et al., 2009).

EDs can be marked by other significant medical problems that arise during a crucial period of adolescent development. These problems include growth retardation, pubertal delay or interruption, peak bone mass reduction, and psychosocial interference. Taken together, both the medical and psychological impact of these illnesses highlight the importance of early intervention to disrupt the course of illness and optimize recovery rates (Treasure & Russell, 2011).

The current review consists of studies completed after the prior review by Lock (2015), which was an update to the original review by Keel and Haedt (2008). This review follows the guidelines in Lock (2015), focusing exclusively on outpatient psychosocial evidence-based practices for children and adolescents (younger than 19 years of age) with EDs, and does not consider adult psychosocial or hospital-based intervention studies.

Methods

The updates in this review are informed by a literature search across PsycInfo, Cochrane, and PubMed. For PsycInfo, we employed the relevant medical subject headings (MeSH terms) “eating disorders,” adding limitations including age (6–12 years) and adolescent (13–18 years); “clinical trial,” and a period from 2015 to 2022 inclusive (this yielded 76 citations before elimination, with zero duplicates identified). For Cochrane, the same qualifiers and limitations were used, yielding 108 citations before elimination. Six duplicates were identified. Lastly, for PubMed, using the aforementioned qualifiers and limitations, 214 citations were accrued before elimination. Thirty-seven duplicates were identified. Across search engines, delimiters, and filters (such as English language only, human subjects, and randomized clinical trial) helped narrow the results, along with the use of Boolean operators (AND, OR, and NOT) to include the following: family therapy, psychopharmacology, comorbid, treatment outcome. In total, 347 articles were eliminated due to being classified as adult studies, single-case reports, inpatient samples, protocol papers, and anything else deemed irrelevant to the scope of this review (e.g., an article about seizures). The search was done by two independent reviews for each search engine, and two meetings were held to review and consolidate the total list of articles. This extra step was taken to ensure that, to the best of our knowledge, all relevant articles were captured and included in this update.

Expert colleagues were consulted to identify any additional studies that may have been missed within this search, which yielded 10 additional articles. For this evidence-based update, two authors systematically reviewed 83 publications for possible inclusion in this manuscript. Of these, 31 articles were deemed to fit the previously described criteria and are included below in Table 1.

Table 1. Outpatient psychosocial treatment studies for child and adolescent eating disorders.

Authors	ED Type	Sample	Treatments	Trial Type	Results
Accurso et al. (2018)	AN	N = 11 adolescents 11–18 years	FBT+DBT skills	Case series	Improvements in weight (large effect size for %mBMI change) and parent-reported eating disorder symptoms but not adolescent-reported; 2 participants met remission criteria at EOT.
Craig et al. (2019)	AN, atypical AN, BN, atypical BN	N = 54 adolescents 13–18 years	CBT-ED	Case series	Improvements in ED cognitions (EDE-Q; large effect size) and clinical impairment levels (moderate effect size) at EOT; AN/atypical AN adolescents had significant increases in %EBW.
Dalle Grave et al. (2015)	BN, BED, or other specified ED	N = 68 adolescents 13–19 years	CBT-E	Case series	Reduced eating disorder psychopathology (EDE-Q; large effect size); decreased binge eating and purging frequency; two-thirds of sample had EDE-Q scores below 1SD of community norms by EOT.
Dalle Grave et al. (2019)	AN	N = 49 adolescents 11–18 years	CBT-E	Case series	Improvements in weight gain (83% IBW to 100.6% at EOT and 100.4% at FU) and decreased clinical impairment and ED cognitions at EOT and maintained at 20-week follow-up.
Eisler et al. (2016)	AN or EDNOS	N = 169 adolescents 13–20 years	FT-AN or MFT-AN	RCT	60% in FT-AN vs 75% in MFT-AN achieved good-intermediate outcomes (weight, mood, ED psychopathology^†.
Fischer and Peterson (2015)	BED, EDNOS	N = 10 adolescents 14–17 years	DBT	Case series	Reduction in binge eating episodes, purging, and disordered eating cognitions (moderate effect sizes) at EOT.
Glashouwer et al. (2018)	AN, BN, EDNOS	N = 51 females 12–18 years	Evaluative conditioning or placebo control	Crossover design	No differences between groups in primary outcome measures of self-reported body satisfaction, weight/shape concern, and self-esteem. Findings do not support the use of evaluative conditioning as an intervention for adolescents with EDs.
Herbrich et al. (2017)	AN	N = 48 adolescents 12–17.11 years	CRT or TAU	RCT	CRT (10 weekly sessions) had a higher completion rate of 79.2% vs. 58.3% in TAU (10 weekly sessions). BMI showed a trend toward significance for CRT. No other differences in clinical or neuropsycho-logical variables.
Herscovici et al. (2017)	AN	N = 23 adolescents 12–20 years	Outpatient Family Therapy with and without Family Meal	Exploratory longitudinal investigation – only one therapist	Both groups improved significantly at EOT and 6-month FU on weight, psychological (SCL90) and ED symptoms (EDI-2), with no difference between groups. The family meal did not appear to have any weight-related benefit. Full weight remission was defined as >95% EBW, and partial was >85%.
Hilbert et al. (2020)	BED	N = 73 females 12–20 years	CBT or waitlist (WL)	RCT	Fewer monthly binge episodes in CBT versus WL; CBT received higher rates of abstinence from binge eating (51% vs. 33%) and remission from BED (57% vs 33%), and lower ED psychopathology.
Hodsoll et al. (2017)	AN	N = 149 adolescents 12–21 years	TAU vs. TAU+ career intervention (ECHO) with telephone guidance or without telephone guidance (3 conditions)	RCT	ECHO groups showed moderate increase in carer skills and reduction in accommodating and enabling behavior. Patients had minor improvement in BMI in ECHO groups vs. not. Little additional benefit in telephone guidance with ECHO vs. not.
Hurst and Zimmer-Gembeck (2019)	AN	N = 21 adolescents 12–17 years	FBT+CBT-P (CBT module on perfectionism)	Case series	Full remission at EOT in 11 of the 21 participants; Increases in weight and decreases in ED symptoms and perfectionism scores; Improvements in perfectionism correlated with improvements in eating disorder symptoms.
Jaite et al. (2020)	AN	N = 50 females 12–18 years	CBT or DBT	RCT	ED symptoms (EDI-2) decreased, and BMI increased significantly from BL to EOT in both treatments (CBT: EDI-2 d = -0.61, BMI d = 1.04; DBT: EDI-2 d = -0.55, BMI d = 0.7). No differences between treatments.
Kamody et al. (2019)	Subthreshold BED	N = 15 adolescents 14–18 years	DBT skills group	Case series	Reduction in binge eating and emotional eating; Three of the six participants that met criteria for BED at BL did not by EOT.
Le Grange et al. (2015)	BN	N = 130 adolescents 12–18 years	FBT-BN or CBT-A	RCT	FBT-BN achieved higher abstinence rates (no binging and purging for 4 weeks prior to EDE assessment) than CBT-A at EOT (39 vs 20%, p = .04) and 6-month FU (44 vs 25% p = .03). No difference at 12-month FU.
Le Grange et al. (2016)	AN	N = 107 adolescents 12–18 years	FBT or PFT	RCT	Remission higher in PFT than FBT (43% vs. 22%, p = .016) at the six month time-point defined in this study as EOT, but not at 6-month FU (12 month time point – the comparative EOT in most other FBT-AN studies; 39 vs. 22%, p = .05) or 12-month FU (37 vs. 29%, p = .44)
Le Grange et al. (2020)	AN, BN, BED, Unspecified Eating Disorder	N = 97 adolescents 12–19 years	FBT or CBT-E	Case series	No difference in ED cognitions between groups at any time; faster weight gain achieved in FBT compared to CBT at EOT only.
Lock et al. (2015)	AN	N = 45 adolescents 12–18 years	FBT or FBT + IPC	RCT	Poor early responders (<2.4 kg by session 4) who received IPC achieved full weight restoration (>95% of expected mean BMI) by EOT at similar rates as those who had responded early.
Lock et al. (2018)	AN + obsessive-compulsive features	N = 30 adolescents 12–18 years	FBT+CRT (Cognitive Remediation Therapy) or FBT+AT (Art Therapy)	RCT	Both treatments rated acceptable; Both groups demonstrated correlational changes in obsessive-compulsive traits and cognitive inefficiency; Greater change in cognitive inefficiencies in FBT+AT group; No difference in clinical outcomes between groups.
Lock et al. (2019)	ARFID	N = 28 children 5–12 years	FBT-ARFID or Usual Care (UC)	RCT	FBT-ARFID had greater change %EBW and weight from BL to EOT compared to UC (large effect size) as well as parental self-efficacy.
Lock et al. (2021)	AN	N = 40 adolescents 12–18 years	GSH-FBT or FBT-V	RCT	Both treatments resulted in improvements in weight, EDE, parental self-efficacy (medium to large ES); Comparable acceptability ratings.
Mazzeo et al. (2016)	LOC-Eating Disorder or BED	N = 45 adolescent females 13–17 years	DBT-based intervention or weight management groups	RCT	No difference between groups; Reductio-ns in eating disorder cognitions, dietary restraint, and eating due to negative affect reported across both groups.
Peterson et al. (2020)	AN, Atypical AN, OSFED	N = 18 adolescents 13–18 years	FBT + DBT skills	Case series	Small to medium effect sizes for decrease in binge eating and eating disorder cognitions and increase in %EBW from BL To EOT.
Philipp et al. (2021)	AN	N = 102 caregivers of adolescents 10–19 years with AN	Supporting Carers of Children and Adolescents with Eating Disorders in Austria (SUCCEAT) – comparing workshop group to online group	Quasi-randomization (allocated to groups alternatively)	No overall difference between groups at EOT; There were similar improvements in BMI, psychopathology, and quality of life for youth with AN across both groups. At 12-month f/up, higher rates of remission in online group (48.4%) compared to workshop group (28.1%) but not statistically significant.
Salerno et al. (2016)	AN	N = 149 dyads parent-adolescent with AN	ECHO (Experienced Carer Helping Others) or Treatment as Usual	RCT	BL distress predicted post-treatment distress, regardless of group; In TAU condition, patient distress at BL related to carer distress at 12-month FU; Carer distress did not predict patient weight change in either condition.
Spettigue et al. (2015)	EDNOS, AN, BN	N = 51 parents/carers and 36 adolescents awaiting ED assessment	Intervention (2-hour initial psychoeducation session with bi-weekly phone calls until assessment) or control (no intervention)	RCT	Intervention group reported higher parental self-efficacy on Parents Versus Anorexia Scale (PVAN) after psychoeducation session and at assessment, and greater knowledge of eating disorders at assessment than control group; No difference in patient outcomes (depression, anxiety, eating disorder cognitions/behaviors, BMI).
Stefini et al. (2017)	BN or partial BN	N = 81 adolescents 14–20 years	CBT or psychodynamic therapy (PDT)	RCT	Remission rates did not differ between groups; Improvements for secondary outcome measures in both groups; Small effect size between group differences on purging and binge eating which favored CBT and eating concern which favored PDT.
J. J. Thomas et al. (2020)	ARFID	N = 20 youth 10–17 years	CBT (CBT-ARFID)	Case series	PARDI (Pica, ARFID, Rumination Disorder Interview) scores improved; Added 16.7 new foods; Average weight gain of 11.5 pounds for underweight subgroup; 70% participants did not meet ARFID criteria at post-treatment.
Timko et al. (2015)	AN or subthreshold AN	N = 47 adolescent 12–18 years	Acceptance-based Separated Family Treatment (ASFT)	Case series	Almost half (48%) of sample met criteria for full remission (95% EBW and EDE scores within 1SD of population norms) at EOT with another third meeting partial remission criteria (90% EBW).
Wade et al. (2022)	AN or atypical AN	N = 16 parents of adolescent 13–15 years on waitlist for FBT	Guided self help family based treatment (GSH-FBT)	Case series	Low uptake (7% of eligible families completed protocol); Improvements in weight, reduced eating disorder behaviors, improvements in mood, and increases in parental knowledge, skills, and confidence.

^†For Eisler et al. (2016), “good outcomes” were defined as patients whose weight was >85% EBW. Of note, this is lower than remission used in the majority of the studies included in this manuscript (Couturier & Lock, 2006).

For this manuscript, remission is defined as >95% EBW with EDE global scores within one SD of published norms, which is 1.73 (Couturier & Lock, 2006). We have highlighted whether studies included in this manuscript used different remission criteria.

Summary of Empirically Supported Psychosocial Interventions

There are 31 additional empirically supported psychosocial interventions for eating disorders published in the last seven years that we identified through the aforementioned literature search (Table 1). A review of these studies is broken down below by disorder subtype.

Anorexia Nervosa

Of these 31 total studies, nine covered review of randomized clinical trials (RCTs) for adolescent AN, spanning 641 adolescents aged 12–20 years (Le Grange et al., 2016; Herbrich et al., 2017; Herscovici et al., 2017; Hodsoll et al., 2017; Jaite et al., 2020; Lock et al., 2021, 2015; Timko et al., 2015). An additional 162 adolescents with AN or atypical AN were included in six case series (Accurso et al., 2018; Dalle Grave et al., 2019; Hurst & Zimmer-Gembeck, 2019; Peterson et al., 2020; Timko et al., 2015; Wade et al., 2022). Two studies examined eating disorders more broadly within the same study, using samples of adolescents with AN or an unspecified eating disorder diagnosis (Eisler et al., 2016) and another case series inclusive of AN, Atypical AN, or OSFED (Peterson et al., 2020). Of note, Eisler et al. (2016) used lower weight-related thresholds in their sample (>85% EBW vs. >95% EBW) to categorize “good outcomes.” Additionally, one quasi-randomization structure intervention study for parents/carers of 102 adolescents aged 10–19 years old with AN was identified (Philipp et al., 2021). According to the previous review, there were 12 completed RCTs of 1,060 adolescents with AN aged 12–20. For this update, we add nine additional RCTs to this number, which are reviewed in additional detail in this manuscript.

No new studies met criteria as a Level 1 (well-established) treatment (Table 2). Currently, the gold-standard treatment for adolescent anorexia nervosa remains family therapy with a behavioral focus, typically referred to as family-based treatment (FBT); (Lock & Le Grange, 2015). Several of the studies included in this review, however, made adjustments to FBT (such as delivery with and without a family meal); (Herscovici et al., 2017), or assessed a parent-only format of FBT (Hodsoll et al., 2017; Salerno et al., 2016) or a guided self-help version (Lock et al., 2021). To that end, a parent-focused treatment (PFT) demonstrated higher remission rates compared to FBT at EOT at six months, but not at 12-and 18-month follow-up time points; thus, the outcome in this study used a different assessment time point than others (which used nine or 12 months), and comparisons to EOT outcomes in other studies need to take this into account. This study fell under a Level 2 (probably efficacious) category (Le Grange et al., 2016). The adaptations of PFT to FBT included having adolescents attend a 15-min session with a nurse to discuss weight and medical stability, and to receive brief supportive counseling, before their parents’ session with the therapist. Results provide preliminary support for the efficacy of a separated model of FBT as an alternative therapeutic platform for rapid weight restoration in adolescent AN.

Table 2. Evidence-based psychosocial interventions for eating disorders in adolescents.

	AN	BN	BED	ARFID
Level 1. Well-Established Treatments	Family-based treatment	Family-based treatment – Bulimia Nervosa	None	None
Level 2. Probably Efficacious Treatments	Family Therapy-Systemic; Insight Oriented-Psychotherapy (individual); Adolescent-Focused Therapy (individual), Parent-Focused therapy		None	None
Level 3. Possibly Efficacious Treatments	GSH, Telephone assisted therapy (Experienced Carer Helping Others, or ECHO); Family-based treatment +Cognitive Remediation Training and Family-based treatment +Art Therapy	Cognitive Behavioral Therapy (Guided Self-Help) (Individual)	Internet Facilitated Cognitive Behavior Therapy (Self Help); CBT; FBT	Family-based treatment
Level 4. Experimental Treatments	Cognitive Behavioral Therapy-Broad (Individual); Cognitive Remediation Training (individual), Family-based treatment-Intensive Parental Coaching; DBT skills group; Family-based treatment + Dialectical Behavioral Therapy Skills; Family-based treatment-Guided Self-Help for waitlist	Cognitive Behavioral Therapy – (Individual); Supportive Psychotherapy (Individual)	Interpersonal Psychotherapy (Individual); Dialectical Behavior Therapy (Individual and Family)	Cognitive Behavioral Therapy
Level 5. Treatments of Questionable Efficacy	Outpatient Family Therapy with/without Family Meal	Guided Self-Help-Waitlist

Guided-self-help therapies and telephone-based treatments fell within the Level 3 (possibly efficacious) categories (Hodsoll et al., 2017; Lock et al., 2021), with both emphasizing treatment delivery/guidance/coaching to caregivers of patients with AN rather than to patients directly. Level 3 (possibly efficacious) treatments also include adjunctive treatments to FBT, such as adding Cognitive Remediation Training (CRT) or Art Therapy (AT) to FBT for adolescents with AN and obsessive-compulsive features (Lock et al., 2018).

Level 4 experimental treatments in this update included studies of Cognitive Remediation Training (CRT) and a feasibility study assessing the efficacy of adding intensive parent coaching to standard FBT (Herbrich et al., 2017). Case series studies investigating CBT-E (Dalle Grave et al., 2019), Acceptance-based Separated Family Therapy (ASFT); (Timko et al., 2015) and GSH-FBT for carers/parents on a waitlist for FBT (Wade et al., 2022) fall within the Level 4 (experimental treatment) category due to the non-randomized study design. Additional adaptations to FBT for adolescent AN (e.g., combining modules and skills from CBT and DBT) are classified as level 4 (experimental treatment) due to limited case series data available at this time (Accurso et al., 2018; Hurst & Zimmer-Gembeck, 2019; Peterson et al., 2020). One study of 21 adolescents that added a CBT-based perfectionism module to FBT (Hurst & Zimmer-Gembeck, 2019) reported correlations between improvements in perfectionism and improvements in eating disorder symptoms, but the limited case series data classifies this FBT treatment adaptation as Level 4 (experimental treatment). Similarly, large effect sizes for percent median BMI change from BL to EOT were reported in a study adding DBT skills to FBT among 11 adolescents with AN (Accurso et al., 2018); although parent-reported eating disorder symptoms improved, adolescent-report of their symptoms did not. Further, only two participants met remission criteria at EOT (Accurso et al., 2018).

Lastly, one exploratory longitudinal study (one therapist across both arms) assessed family therapy conducted with and without a family meal; the limitations of this exploratory study design make it a Level 5 (treatment of questionable efficacy) category (Herscovici et al., 2017).

Bulimia Nervosa

At the time of the previously published reviews (Keel & Haedt, 2008; Lock, 2015) only two published RCTs involving 165 adolescent participants existed. These studies examined FBT for BN, a guided self-help version of CBT, and individual supportive therapy (Le Grange et al., 2007; Schmidt et al., 2007). Since the earlier review, only two additional RCTs were conducted examining psychosocial interventions for BN, spanning an additional 211 adolescents aged 12–20. These studies examined FBT-BN or CBT-A, finding FBT-BN superior in achieving remission rates (Level 1, well established); (Le Grange et al., 2015) and CBT or psychodynamic therapy (PDT), finding no significant differences between the two groups on remission rates (Level 3, possibly efficacious); (Stefini et al., 2017). This second RCT on FBT-BN by Le Grange and colleagues in 2015 found it superior to comparative treatments, thus making FBT-BN a Level 1 treatment, which is an improvement from its categorization in the prior review. This makes FBT the only Level 1 RCT for BN adolescents to our knowledge. Compared to adult literature, relatively little is known about treatments for BN in adolescents.

Binge Eating Disorder

RCTs for adult BED have gained considerable attention, finding CBT and IPT useful (Wilson et al., 2010), and showing preliminary support for DBT (Safer et al., 2010). In adolescents with BED, the prior review reports the use of IPT having preliminary success along with internet-delivery of CBT self-help (Jones et al., 2008; Tanofsky-Kraff et al., 2010). The current review adds two RCTs to this relatively small pool of trials, spanning an additional 118 youth with BED aged 12–20. Hilbert et al. (2020) found CBT for adolescent BN superior to a waitlist comparison group; this study classifies as a Level 3 (possibly efficacious treatment). Mazzeo et al. (2016) found both DBT-based intervention and a weight management intervention for adolescents with LOC-ED or BED reduced disordered eating cognitions, suggesting these treatments may be possibly efficacious (Level 3). However, the outcomes regarding reduction in binge eating episodes for either treatment group were not reported (Mazzeo et al., 2016); thus, it is hard to determine what level of efficacy these interventions might have for binge eating episodes specifically. Two case series using DBT for BED or subthreshold BED were identified (Fischer & Peterson, 2015; Kamody et al., 2019), both Level 4 (experimental treatments); these studies jointly noted reductions in binge eating episodes by EOT. Fischer and Peterson (2015) reported moderate effect size estimates for reductions in disordered eating cognitions, and Kamody et al. (2019) observed decreases in self-reported emotional eating. Among the six participants that met full criteria for BED at BL, three no longer met criteria following the DBT skills group intervention at EOT (Kamody et al., 2019). However, these data obtained from small samples limit generalizability, and larger scale randomized controlled trials are needed to better evaluate the role of DBT in adolescent BED.

Avoidant Restrictive Food Intake Disorder

Research on treatments for youth with ARFID is in its infancy. Very few studies have examined psychosocial treatments for ARFID in children and adolescents. Only one ARFID RCT (Lock et al., 2019) fell at a Level 3 (Possibly Efficacious). There was also one case series (J. J. Thomas et al., 2020) that fell within a Level 4 (Experimental Treatments). In total, both these preliminary studies encompassed 48 youth aged 5–17 years old. No ARFID studies were identified in the prior review (Lock, 2015). Although outcome data from randomized clinical trials are limited, published treatment manuals are available for both CBT-ARFID (Thomas & Eddy, 2019) and FBT-ARFID (Lock, 2022).

Transdiagnostic

Three case series studies investigating CBT-E included participants with eating disorders or sub-threshold eating disorder symptoms across diagnostic categories (Craig et al., 2019; Dalle Grave et al., 2015; Le Grange et al., 2020). Although these studies include a substantial number of youths with a range of reported eating disorder diagnoses and symptoms (n = 219), these studies all fall under Level 4 (experimental treatment), as they have not been tested in an RCT. Further, one study allowed adolescents/families to choose which treatment option they would prefer (FBT or CBT-E), which could impact intervention outcomes and findings (Le Grange et al., 2020). Although no differences were observed in ED cognitions between the groups at any time point, by EOT only, participants in FBT gained weight at a faster rate than participants in CBT (Le Grange et al., 2020). Further, the authors found that older patients and those with greater eating disorder symptom severity chose CBT-E more often than FBT, suggesting these factors could be important moderators to consider in clinical settings where patients are requesting specific treatment modalities. Another study using a crossover design randomized adolescent females with AN, BN, and an unspecified eating disorder diagnosis to a placebo-control or an experimental condition using a computer-based evaluative conditioning intervention. There were no differences between outcome measures of body dissatisfaction, shape/weight concern, or self-esteem between conditions at the end of treatment and at 3- and 11-week follow-up time points (Glashouwer et al., 2018).

We also identified a study for parents/carers of patients waiting to be assessed for an eating disorder (Spettigue et al., 2015). In this RCT, authors compared a two-hour initial psychoeducation session plus bi-weekly phone calls until assessment compared to a control (no intervention) condition and found that the intervention group reported higher parental self-efficacy and greater knowledge of eating disorders than the control group, although no differences in patient outcomes, such as depression, anxiety, BMI, and eating disorder cognitions, were observed. Because of high wait times, studies of interventions to assist parents and children are an important future research area.

Highlighted New Studies [Since 2015]

Several studies included in the current review encompass feasibility studies, setting the stage for larger, adequately powered treatment studies. For instance, the ARFID pilot done by Lock et al. (2019) demonstrated a greater change in %EBW for the FBT cohort from BL to EOT relative to the UC; this difference was accompanied by a large effect size. Informed by these novel findings, a larger RCT is currently underway. Additionally, given ARFID’s nascent introduction to the DSM-5, a case series investigating treatment efficacy by J. J. Thomas et al. (2020) found CBT for ARFID useful – specifically, symptom improvement and weight gain. They report that 70% of participants did not meet ARFID criteria at a post-treatment time point. A feasibility study demonstrating acceptability and identifying potential mechanisms would be a logical next step to see if this approach warrants further study using a fully powered RCT.

In the context of the COVID-19 pandemic, there were several treatment efforts pivoting care to virtual delivery. This has been both a necessary and innovative way to meet the needs of ED patients during a time of crisis, and has led to several trials adapting known, evidence-based treatments (such as FBT and CBT) to various virtual modalities (virtual delivery versus a guided self-help modality). The benefits of telehealth modalities include ease of access and decreased burden related to travel for carers or parents (Hodsoll et al., 2017; Lock et al., 2021). Initial data is promising, revealing that the use of teleconferencing is feasible and that its treatment effects appear similar to in-person intervention.

Moderators and Mediators of Treatment Effect for Level 1 and 2 Studies

Few additional studies have added to a systematic evaluation of moderators and mediators of treatment effects for eating disorders in youth. Some previous studies have explored the role of expressed emotion (parental criticism) on treatment outcomes in FBT for AN (Rienecke, 2017; Rienecke et al., 2021); however, high parental expressed emotion at baseline did not predict negative outcomes four years later. A secondary data analysis found that driven exercise at BL predicts worse outcomes (greater ED severity) for adolescents with AN but not adolescents with BN at EOT (Eisler et al., 2000; Stiles-Shields et al., 2015), but this was a nonspecific predictor not associated with treatment type.

In a separate investigation of mediators, parental self-efficacy was examined as a potential mechanism for change, particularly early response (or weight gain of at least 2.4 kg within the first four sessions) in FBT for AN (White et al., 2017). Here, direct and non-direct parent eating prompts during family meal sessions in FBT for AN resulted in more weight gain at EOT, though this was not associated with disordered eating severity in the adolescent. Understanding mechanisms is an important area for continued research but is hampered by few RCTs with sufficient numbers to allow for mediator assessment (Kraemer et al., 2002).

Emerging Treatments

As cited in the prior review by Lock (2015), recovery amongst adolescent eating disorders remains 30–40% despite the rise in RCTs and efforts to identify effective interventions. This highlights the ongoing need to both optimize existing treatments as well as identify novel treatments targeting EDs. One promising transdiagnostic approach includes guided self-help modalities. Such modalities are not only useful for carers, but they also minimize therapist time and streamline the delivery of salient treatment materials. Lock et al. (2021) provided preliminary data (included in this review) demonstrating that a guided self-help modality of FBT (including 20–30 min guided self-help sessions with parents or carers) had comparable outcomes (improvements in weight, cognitions, and parental self-efficacy) to FBT delivered virtually (FBT-V). A follow-up mixed methods study looking at parent and clinician perspectives within this RCT found that parents reported improvement in their child’s AN symptoms regardless of treatment condition, while clinicians reported lower competency and comfort metrics within the guided self-help modality versus FBT (Matheson et al., 2022). In a case series done by Wade et al. (2022), GSH-FBT for families on a waitlist for ED treatment demonstrated improvements in mood, weight, and ED behaviors – highlighting the important utility of GSH modalities to help the increasing need for treatment nationally. Waitlists have continued to grow during the COVID-19 pandemic, and jointly, these preliminary studies provide hope for offering evidence-based treatments widely at reduced clinician time and cost.

In addition to guided self-help modalities, the pandemic’s need to transition services to telehealth has led to a rise in telehealth treatments. Indeed, virtual delivery of treatments has been found to have comparable clinical outcomes to in-person services; recent research has helpfully outlined how to effectively pivot gold-standard treatments to video-conferencing (Matheson et al., 2020). Some research has also highlighted the benefits of virtual delivery, including reduction of travel time, ability to access more remote areas that are not near academic medical centers, and greater insight into patient’s lives (Smith et al., 2020). Indeed, the prior review predicted emerging treatments “will likely evolve utilizing internet or phone application supported/delivered treatments,” and we are certainly noticing a rise in such treatments both in the context of the pandemic and finding their utility in its aftermath. Further developments in using phone messaging and app interfaces may be beneficial in providing real-time feedback of skills in the moment, for example, delivering emotion regulation during times of distress for ED patients.

Lastly, there has been a notable rise in ARFID treatment development, given ARFID’s relatively new inclusion in the DSM-5. Over the past decade, ARFID mechanistic and treatment research has been of great interest. Several RCTs are currently underway to help treat ARFID symptoms in youth. Included in this review are both a case series and feasibility RCT for ARFID, adapting two known, evidence-based treatments for ARFID presentations – CBT and FBT, respectively (J. J. Thomas et al., 2020; Lock et al., 2019). While case studies were not included in this review, Zucker et al. (2019) mentions the adaptation of a prior RCT for youth with functional abdominal pain for children with ARFID – The “Feeling and Body Investigators.” This case series describes the utility of an interoceptive-exposure-based treatment for a four-year-old with ARFID and may necessitate larger-scale investigation.

Limitations

Lock (2015) summarizes limitations in the prior review, noting some improvements in the 2008–2015 timeframe from limitations addressed in the original manuscript (Keel & Haedt, 2008). These included expanding samples to include males, increased racial and ethnic diversity, and generally increased number of child and adolescent RCTs broadly in EDs (specifically, the number of adolescents included in RCTs was reported as tripling from 302 in 2008 to 1060 in 2014). This number has continued to grow in the past seven years, with the addition of 21 RCTs. Male participants and more diverse samples have been included in these trials, but there is still a large skew toward Caucasian females in ED RCTs transdiagnostically. Additionally, the criticism from Lock (2015) regarding treatments for AN still stands: “Large studies have examined only two types of family therapy (family-based treatment and systemic family therapy) and one type of individual therapy (adolescent focused therapy), whereas other potentially effective interventions have a more limited database of support in this age group.” However, one notable improvement over the past seven years is the inclusion of telehealth or guided-self-help modalities for AN and BN alike. Potentially a focus of interest in the context of the COVID-19 pandemic, virtual treatment delivery, and guided self-help modalities have become both a necessity and a convenient way of relaying salient treatment information to vulnerable groups and their carers.

Recommendations for Best Practice

To echo conclusions drawn from both the prior reviews of the literature (Keel & Haedt, 2008; Lock, 2015), FBT for adolescents with AN and BN remains the only treatment meeting the “well-established treatment criteria.” These conclusions continue to be supported by case series data finding faster weight gain in FBT relative to CBT-E at EOT (Le Grange et al., 2020). Of note, several studies included in this review have added adjunctive treatments to FBT (such as FBT+ DBT skills, or FBT + adaptive interventions for those who do not demonstrate early response), potentially optimizing and tailoring the treatment to different presentations of AN. Additionally, efforts to optimize FBT treatment delivery and access have been underway and discussed in prior sections, such as adaptive and guided self-help versions of FBT. Future investigation of how to integrate technology (both phone and app-based delivery) may be useful to continue to optimize access to gold-standard treatments while reducing cost.

Additionally, this review includes several case series studies that report improvements in clinical outcomes accompanied by medium-to-large effect sizes (Craig et al., 2019; Dalle Grave et al., 2015; Fischer & Peterson, 2015; Peterson et al., 2020). These provide important data on the potential utility of these interventions, though larger-scale studies are needed for generalizability and replication. The use of different diagnostic criteria for inclusion/exclusion as well as remission criteria (e.g., weight cutoffs, cognition) can make it challenging to interpret intervention results across samples (Wade & Lock, 2020). To that end, there is an additional need for studies to look specifically at unspecified eating disorders or subclinical categories (Dalle Grave et al., 2015). These groups often go unstudied due to not “neatly” falling into a diagnostic category but may still well benefit from interventions.

There is a need to increase diversity in samples to ensure both cultural sensitivity and efficacy across populations – including, but not limited to, racial and ethnic minorities, LGBTQ+, and transgender populations. A recent call to action paper suggests actionable steps to increase the diversity of underrepresented racial and ethnic groups to address justice, equity, diversity, and inclusion in the ED field (Goel et al., 2022). Some of these steps include community-engaged practices and purposeful sampling, employing culturally sensitive language during recruitment and consent, and using culturally appropriate measures that have been normed on the population of interest. Future studies may benefit from consideration of these steps to improve representation of historically excluded racial/ethnic populations in research.

Disclosure Statement

James D. Lock has the following commercial relationship(s) to disclose: National Institute of Mental Health for research funding, ownership interest in the Training Institute for Child and Adolescent Eating Disorders, royalties from Oxford Press, Guilford Press, Taylor & Francis, Routledge, and American Association Press.