Multi‐family therapy for eating disorders: A systematic scoping review of the quantitative and qualitative findings

Abstract Objective This study reviewed the quantitative and qualitative evidence‐base for multi‐family therapy (MFT) for eating disorders regarding change in physical and psychological symptoms, broader individual and family factors, and the experience of treatment. Method A systematic scoping review was conducted. Four databases (PsycInfo, Medline, Embase, CENTRAL) and five grey literature databases were searched on 24th June 2021 for relevant peer‐reviewed journal articles, book chapters, and dissertations. No beginning time‐point was specified. Only papers that presented quantitative or qualitative data were included. No restrictions on age or diagnosis were imposed. Studies were first mapped by study design, participant age, and treatment setting, then narratively synthesized. Results Outcomes for 714 people who received MFT across 27 studies (one mixed‐method, 17 quantitative and nine qualitative) were synthesized. MFT is associated with improvements in eating disorder symptomatology and weight gain for those who are underweight. It is also associated with improvements in other individual and family factors including comorbidities, self‐esteem, quality of life, and some aspects of the experience of caregiving, although these findings are more mixed. MFT is generally experienced as both helpful and challenging due to the content addressed and intensive group process. Discussion MFT is associated with significant improvements in eating disorder symptoms across the lifespan and improvement in broader individual and family factors. The evidence base is small and studies are generally underpowered. Larger, higher‐quality studies are needed, as is research investigating the unique contribution of MFT on outcomes, given it is typically an adjunctive treatment.

occurs in various constellation, such as separate young person, sibling and parent groups, mixed groups, or pairs, etc. The group will usually also eat up to three meals together during each MFT day. MFT-AN is now a recommended treatment for adolescents by several practice guidelines (Couturier et al., 2020;Heruc et al., 2020;NICE, 2017) and a specific version has recently been developed for adolescents with bulimia nervosa (MFT-BN; Stewart et al., 2019).
All MFT models are designed to improve treatment outcomes by reducing perceived isolation and stigma, enhance family relationships and promote family skill building (Asen & Scholz, 2010;Dawson, Baudinet, Tay, & Wallis, 2018;. Some models also specifically aim to intensify treatment, particularly at the early stages Wierenga et al., 2018), which has been shown to be a critical time during treatment. Early eating disorder symptom change has been shown to be a robust predictor of end of treatment outcomes across diagnosis, age range, treatment type, and setting (Nazar et al., 2017;Vall & Wade, 2015); hence, the importance of a more intensive intervention, such as MFT, at this stage of treatment.
Given MFT can provide early, intensive support, that focuses on both patient and family factors, it has great potential to improve upon current treatment outcomes either as a stand-alone or adjunctive intervention. Its use also fits with practice guidelines, which increasingly suggest involving family members in child, adolescent, and adult treatments (Fleming, Le Brocque, & Healy, 2020;Hilbert, Hoek, & Schmidt, 2017; National Institute for Health and Care Excellence (NICE), 2017; Treasure, Parker, Oyeleye, & Harrison, 2021).
Emerging evidence indicates that MFT is associated with improved physical health, a reduction in eating disorder symptoms and improvements in a range of other patient and family factors, such as selfesteem, quality of life, and caregiver burden (Gelin et al., 2016). Results from the only outpatient randomized controlled trial (RCT) published indicate global outcomes at discharge from family therapy with adjunctive MFT are improved compared to family therapy alone for adolescents with anorexia nervosa (Eisler, Simic, Hodsoll, et al., 2016).
Yet, despite its promise, MFT remains relatively understudied. The heterogeneity of MFT models described and evaluated, as well as the relatively high resource cost and intensity required of some MFT models, makes it difficult to implement and evaluate. Furthermore, MFT is rarely a stand-alone treatment and large variability exists between studies in the way MFT is delivered, including setting (inpatient, day program, outpatient), treatment duration, and treatment intensity (Gelin et al., 2018), making MFT-specific findings difficult to generalize.
Given MFT is now widely used in clinical services internationally, a systematic scoping and synthesis of the available data is needed to determine the evidence base and identify gaps for future research. To better understand the impact of MFT on eating disorder treatment outcomes, this study aimed to systematically review and synthesize the available quantitative and qualitative findings. While a review has previously been completed of MFT for a range of psychiatric disorders (Gelin et al., 2018), including eating disorders, this was not exhaustive, missed some important papers (Jewell & Lemmens, 2018), and did not include qualitative data. Specifically, this review has three aims: 1. To review the impact of MFT on the physical and psychological symptoms of eating disorders.
2. To review the impact of MFT on families and caregivers.
3. To review the individual and family experience of receiving MFT.

| METHOD
A systematic scoping review methodology (Peters et al., 2015) was used to explore the existing research into MFT for eating disorders across the age range. This was identified as the most appropriate methodology given the heterogeneity of existing research and the broad aims of this review. This allowed for more descriptive studies that included some outcome data to be included. Current scoping review guidelines (Peters et al., 2020) and Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) extension for scoping reviews guidance (Tricco et al., 2018) were used to conduct this review. The research was reviewed and approved by an institutional review board.
The methodology was initially developed by one author (JB) using the PICOS (population, intervention, comparison, outcome, study design) framework (Methley, Campbell, Chew-Graham, McNally, & Cheraghi-Sohi, 2014). Two authors (JB, LD) then independently executed the search strategy, study selection and data extraction. Disagreements were resolved by consensus discussions. Data were reviewed using a parallel-results convergent synthesis design (Noyes et al., 2019), whereby quantitative and qualitative data were initially analyzed and presented separately; then, synthesized for interpretation of the findings. This was deemed the most appropriate method of initially scoping both the quantitative and qualitative data, as well as synthesizing all available data.

| Eligibility criteria
Eligibility criteria for this review are presented in Table 1 Abbreviations: ACT, acceptance and commitment therapy; AN, anorexia nervosa; AN-rd, anorexia nervosa and related disorders; ARFID, avoidant/restrictive food intake disorder; Ax, assessment; BED, binge eating disorder; BMI, body mass index; BN, bulimia nervosa; BN-rd, bulimia nervosa and related disorders; CBT, cognitive behavioral therapy; CRT, cognitive remediation therapy; DBT, dialectical behavior therapy; DP, day program; ED, eating disorder; ED-Rs, restrictive eating disorders; EDNOS, eating disorder not otherwise specified; EDNOS-R, eating disorder not otherwise specified characterised by restriction; EOT, end of treatment; FBT, family-based treatment; FT-AN, family therapy for anorexia nervosa; FU, follow up; IBW, ideal body weight; IOP, intensive outpatient program; IP, inpatient; MDT, multi-disciplinary team; MI, motivational interviewing; OSFED, other specified feeding and eating disorder; OSFED-R, other specified feeding and eating disorder characterized by restriction; PG, parent group; PHP, partial-hospitalization program; PMM, predictors, moderators or mediators; RO DBT, radically open dialectical behavior therapy; SES, socioeconomic status; UFED, unspecified feeding and eating disorder. a Significant testing compares baseline to follow-up period. b Median reported instead of mean. Full remission was defined as normal weight (≥95% of expected for sex, age, and height), Eating Disorder Examination Questionnaire (EDE-Q) global score within 1 SD of norms, and absence of binge-purging behaviors. Partial remission was defined as weight ≥85% of expected or ≥95% but with elevated EDE-Q global score and presence of binge-purging symptoms (<1/week). c Definition of full remission: at least 95%EBW and EDE global score within 1 SD of community norms (Lock, 2018). IIP: means, SD nr SASB-Intrex: means, SD nr "Less domineering" ns "Less vindictive"* "Less self-blaming" ns "More controlling to mo." ns "More submissive to mo." ns "More emancipating to fa." ns "Fa. more ignorant and distant" ns Note: When data for two treatment groups are reported (e.g., MFT and FT) significance values reported are for time effect, not treatment or interaction effect.

| Selection process
After completing the initial search, duplicates were deleted, and the remaining titles and abstracts were reviewed. Full-text citations and reference lists for relevant articles were screened for eligibility before reaching consensus at the included papers in this synthesis (see Figure 1 for PRISMA flowchart). Zotero software was used in this process.

| Study selection and characteristics
Nine-hundred-and-seventy-two papers were initially identified through the systematic literature search. After duplicates were deleted and screening was performed according to the eligibility criteria (Table 1), a total of 27 articles were determined eligible for this review (see Figure 1 for PRISMA flowchart). The total sample reported on who received MFT is 714 (mean age = 18.7 years, range = 11-62, 97% female).
Outcomes from the 27 studies are synthesized below, comprising data generated from one mixed-method, 17 quantitative, and nine qualitative studies. Three of the quantitative studies also included some qualitative feedback data; however, the data analysis methodology was not adequately reported on to reach the review inclusion criteria. As such, only the quantitative data from these studies are included in this review (Dimitropoulos, Farquhar, Freeman, Colton, & Olmsted, 2015;Mehl, Tomanová, Kuběna, & Papežová, 2013;Wierenga et al., 2018). The only included mixed-method study was a doctoral dissertation (Salaminiou, 2005), of which most (but not all) of the outcome data were published in a peer-review journal (Salaminiou, Campbell, Simic, Kuipers, & Eisler, 2017). Both the dissertation and article were identified by the search strategy and included in this review. Data reported in Salaminiou et al. (2017) are reported as such. All remaining quantitative and qualitative data are reported as Salaminiou (2005) henceforth.
Most studies were from Europe (n = 18, 67%) and had relatively small sample sizes. Nearly, a quarter of the quantitative studies had 30 participants or less (n = 6, 22%) and only two (7%) had a sample size greater than 100. Twenty-three studies reported on MFT in an outpatient setting (17 young person, six adult) and four on a day-or inpatient setting (two young person, two adult). Seven studies compared MFT outcomes to another treatment (five young person, two adult). See Table 2 for a summary of included study characteristics.
See Tables 3, 4 and 5 for a summary of the quantitative eating disorder outcomes, quantitative comorbid and family outcomes, and qualitative outcomes, respectively.

| MFT models: Population, setting, intensity, and duration
There was substantial variability in the different types of MFT models described (see Tables 2 and 4). However, when studies were clustered according to age, diagnosis, and setting, more homogeneity emerged.
One commonality between most studies was that MFT was an adjunctive treatment. Apart from a stand-alone 5-day MFT program described in three studies, all from the same center, MFT was always offered in combination with another form of outpatient treatment (e.g., single-family therapy) or as part of an inpatient admission.
Outpatient MFT-AN typically lasted 9-12 months and included between 8 and 21 days of MFT treatment. The only exceptions were the three studies from the same center that offered the stand-alone 5-day MFT-AN model Marzola et al., 2015;Wierenga et al., 2018). Almost all were influenced by the Maudsley Hospital  and/or Dresden (Scholz & Asen, 2001) models. Outcomes for MFT-BN were only described in two studies from the same child and adolescent service, which lasted four months (Duarte, 2012;Stewart et al., 2019). Inpatient MFT models for young people were briefer, ranging from 2 (Depestele et al., 2017) to 4 months (Geist et al., 2000). During the latter program, all participants were discharged to outpatient treatment during the course of MFT (Geist et al., 2000).
Adult MFT models were generally much more varied compared to those described for young people. Outpatient adult MFT included a stand-alone 5-day model (Wierenga et al., 2018), a 26-week model (Tantillo et al., 2019), and a 12-month model (Skarbø & Balmbra, 2020), with the latter specifically targeting a mixeddiagnostic group of adults with "severe eating disorders." Two studies described MFT as part of inpatient or day-patient treatment. One program described was a very brief (3-day) MFT workshop (Whitney, Currin, et al., 2012;Whitney, Murphy, et al., 2012). The other used an 8-week program that was offered to those on the inpatient and daypatient units (Dimitropoulos et al., 2015). Brief MFT was considered more cost effective than family therapy on an adult inpatient unit in one study (Whitney, Murphy, et al., 2012). There were no specific MFT-BN program for adults identified by the search strategy.

Randomized controlled trials
Three studies used an RCT design, all of which compared MFT to a version of single-family therapy (see Table 2). One investigated MFT-AN for young people in an outpatient setting (Eisler, Simic, Hodsoll, et al., 2016), one for MFT for young people on an inpatient unit (Geist et al., 2000), and one for adult inpatient MFT (Whitney, Murphy, et al., 2012). See Table 2 for details.
The largest study (N = 167), and only multi-center trial identified by the search strategy, randomized young people (age range = 12-20) to 12 months of outpatient FT-AN alone or FT-AN plus 10 days of MFT-AN (Eisler, Simic, Hodsoll, et al., 2016). No significant differences between groups at baseline were observed. Regardless of the treatment arm, significant improvements in global outcomes, weight, eating disorder psychopathology, and mood, as well as the negative aspects of caregiving were reported. Participants randomized to receive MFT-AN also had better global outcomes at end of treatment, using the Morgan Russel outcome criteria (Russell, Szmukler, Dare, & Eisler, 1987), compared to those who received FT-AN alone. Seventy-six percent had a Good or Intermediate outcome in the MFT group compared to 58% in the FT-AN group (Eisler, Simic, Hodsoll, et al., 2016). This difference was no longer statistically significant at 6-month follow-up (18-months post randomization); however, the MFT group continued to have significantly higher %mBMI (MFT-AN group = 91% vs. FT-AN group = 85%, respectively). Self-report self-esteem did not change between baseline and end of treatment in either study arm, although the authors note that baseline scores were within the normal range, suggesting a ceiling effect (Eisler, Simic, Hodsoll, et al., 2016).
The remaining two RCTs identified were both much smaller and conducted on inpatient units. Geist et al. (2000) randomized adolescents (N = 25) to receive either single-family therapy or MFT as part of their inpatient treatment package. No baseline differences between the groups were reported. Treatment in both arms was associated with physical health and eating disorder symptom improvement; however, no differences were reported in weight, eating disorder symptoms, or family functioning outcomes between the two treatments (Geist et al., 2000). Contrary to findings from the Eisler, Simic, Hodsoll, et al. (2016) RCT, no changes in symptoms of depression or severity of general psychopathology were reported (Geist et al., 2000). Notably, self-report family functioning significantly deteriorated in both treatments, indicating an acknowledgement of more family psychopathology at the end of treatment.
On an adult inpatient unit, Whitney, Murphy, et al. (2012) randomized participants (N = 48) to either 18 hours of weekly/fortnightly single-family therapy or a 3-day MFT intervention during their admission.
They reported a significant treatment by time interaction effect. Post hoc comparisons showed that participants who received MFT had higher BMI at 6-month follow-up, but lower BMI at 36-month follow-up. However, these did not reach statistical significance. Across both treatments, a significant reduction in expressed emotion and improvement in caregiver general wellbeing was also observed; however, neither the negative nor positive aspects of caregiving significantly changed. Furthermore, no differences between the treatments were reported on any other individual or family outcome measure at short (3-month) and long (3-year) term follow-up, potentially emphasizing the general benefits of family involvement, rather than any MFT-specific benefits.

Non-randomized comparison studies
Two outpatient (both young person) and two inpatient studies (one young person, one adult) compared outcomes following MFT to another type of treatment using a non-randomized design (see Table 2 for details).
In a retrospective chart review of treatment response in a specialist child and adolescent service (N = 50), Gabel et al. (2014)  although these changes were not compared to the control group outcomes. There were no differences between the two groups at baseline. Marzola et al. (2015) also conducted a retrospective chart review.
They compared outcomes at follow-up for two different 5-day versions of intensive outpatient family therapy for young people; intensive single-family, and MFT. End-of-treatment (5 days) outcomes were not compared; however, each is associated with eating disorder symptom improvements and reported separately elsewhere Rockwell, Boutelle, Trunko, Jacobs, & Kaye, 2011). At follow-up (mean = 30.9 months, SD = 20.2) both treatments continue to be associated with improvements, although no differences between groups with respect to %mBMI, global outcomes and need for higher levels of care (inpatient or residential) are reported (Marzola et al., 2015). Of note, the MFT group were significantly younger ( and caregiver factors (perceived burden, expressed emotion, and caregiver symptoms of depression) all significantly improved. These changes were either maintained (caregiver burden) or continued to improve (expressed emotion and caregiver depressive symptoms) at a 3-month follow-up. Nevertheless, the level of perceived social support and impact of stigma for caregivers did not change across treatment or follow-up period (Dimitropoulos et al., 2015). Furthermore, no differences between interventions were reported on any individual (BMI and eating disorder psychopathology) or caregiver factors (perceived burden, expressed emotion, perceived social supports, and stigma) (Dimitropoulos et al., 2015).

Case series
Physical health and eating disorder symptomatology. MFT-AN was associated with significant improvements in weight, regardless of age or treatment setting (see Table 2). Only one study did not report a significant improvement in weight during MFT; however, participants in this study started treatment within the healthy range (mean BMI = 20.7, SD = 3.3), which was maintained during treatment (Tantillo et al., 2019). Significant improvements were also reported in eating disorder psychopathology, including binge-purge symptoms, by every study that measured it, irrespective of age, setting or instrument used (see Table 2).
Comorbid symptomatology. Outpatient MFT-AN for young people was associated with a significant reduction in symptoms of depression from baseline to discharge (Salaminiou et al., 2017). Salaminiou et al. (2017) reported that symptoms of depression reduced from just below the "severely depressed" range to within the "mild" range after 6 months of MFT-AN. Similarly, self-report symptoms of both depression and anxiety significantly reduced during outpatient MFT-BN, although, parent reports of their child's symptoms did not reveal significant changes (Stewart et al., 2019).
The only adult MFT case series to investigate comorbid symptoms found that state, but not trait, anxiety reduced during a 5-day MFT week (Wierenga et al., 2018). Change in symptoms of depression was not investigated in any adult MFT study in this review.
Broader individual functioning and well-being. Several studies assessed broader symptoms of general well-being in addition to eating disorder symptom change (see Table 3). Outpatient MFT-AN for young people was associated with significant improvements in quality of life (Gelin et al., 2015;Mehl et al., 2013), self-perception and self-image (Hollesen et al., 2013), and self-esteem (Mehl et al., 2013;Salaminiou et al., 2017). MFT-BN for adolescents was associated with significant improvements in emotion regulation capacity (Stewart et al., 2019).
Regarding MFT for adults, at the end of treatment, patients reported significant improvements in emotional awareness but no change in emotion regulation strategies (Tantillo et al., 2019). In the inpatient context, difficulties with interpersonal functioning did not change from baseline to short-and long-term follow-up (Whitney, Murphy, et al., 2012).
Family functioning. In outpatient MFT-AN for young people, Salaminiou (2005) found that family functioning did not change during 6 months of MFT, although the author noted that mean scores at baseline were mid-ranged, indicating adequate family functioning. For adults who attend a 5-day MFT, a significant improvement in general family functioning was reported (Wierenga et al., 2018). See Table 3 for further details.
Parent and caregiver factors. Outpatient MFT-AN for young people was associated with a range of caregiver/parent improvements. By the end of treatment, caregiver burden and most negative impacts of the illness significantly reduced in one study (Dennhag et al., 2019).
Perceived caregiver isolation was the only aspect that did not change during MFT (Dennhag et al., 2019). In another study, parental mood improved (Salaminiou et al., 2017). This improvement was significant for mothers, but not fathers, however, baseline maternal and paternal scores were within the normal range, suggesting a floor effect (Salaminiou et al., 2017). In the same study, adjusted regression analysis revealed change in parental depressive symptoms across treatment was not associated with young person percentage median Body Mass Index (%mBMI) outcome at end of treatment (Salaminiou, 2005). Following outpatient MFT-BN, caregiver burden and parental mood also significantly improved in one study, although level of anxiety did not (Stewart et al., 2019).
The impact of MFT on expressed emotion (critical comments, positive remarks emotional overinvolvement, warmth, and hostility) is mixed. Paternal critical comments significantly reduced from baseline to 6 months in one study (Salaminiou, 2005); however, all other aspects of maternal and paternal expressed emotion towards the child did not change during 6 months of treatment. Salaminiou (2005) also measured level of expressed emotion between parents, as one marker of how well the parental dyad was functioning. Again, no change was observed, except for warmth from mothers towards fathers, which significantly increased (Salaminiou, 2005). Furthermore, adjusted regression analysis revealed that a reduction in paternal criticism and an increase in emotional overinvolvement during MFT-AN was associated with improved young person %mBMI at end of treatment in the same study (Salaminiou, 2005).  Table 2 for further details). In addition, emotion regulation capacity was either maintained or continued to improve during the follow-up period (Tantillo et al., 2019). One case series with young people reported that weight continued to significantly improve, while the quality of life and eating disorder symptoms stabilized at 1-year follow up; however, no data were reported (Gelin et al., 2015).
Non-completion rates (all study designs). MFT non-completion, also referred to as dropout, is typically reported to be low and is an oftenstated benefit of the treatment (Gelin et al., 2018). Twelve of the 17 quantitative studies reported dropout rates, which ranged from 0% to 17%. Six of these studies reported dropout rates below 10% (see Table 3 for details). Data are presented here synthesized, rather than by study design, to provide a better overview of all available data.

| Experience of MFT: Qualitative data
Qualitative data are reported in 10 studies: seven reporting on the experience of MFT for young people and their family members, and three on MFT for adults and their family members. Most commonly, data were generated from individual or focus group interviews and responses analyzed using thematic or content analysis. See Table 3 for further details. The total sample reported on consisted of 47 people with eating disorders (30 young people and 17 adults) and 140 caregivers (120 parents, 14 siblings, three partners, two grandparents, and one adult patient's child). Several papers also reported on clinician experience of MFT (Brinchmann et al., 2019;Wierenga et al., 2018;Wiseman et al., 2019aWiseman et al., , 2019b, which is not reported here as it is beyond the scope of this review. Of note, the majority of qualitative data are generated from the family and caregiver perspective. Only six studies (four young persons, two adults) included patients in their sample, and two studies, which appear to use the same sample, noted that they attempted to recruit young people but all declined (Wiseman et al., 2019a(Wiseman et al., , 2019b. Qualitative studies were initially reviewed separately according to MFT target population (young person or adult) and setting (outpatient or day/inpatient) and were intended to be presented separately. However, due to large overlap in participants' experiences, the data are synthesized and presented together.
Across all studies, there was a common finding that MFT is experienced as both helpful and challenging with similar experiences described for adults and young people for both MFT-AN and MFT-BN. From data generated through observation, interviews and focus groups collected during and after treatment, there was a sense by most participants that MFT helped the family to view the eating disorder symptoms in new ways (Baumas et al., 2021;Duarte, 2012;Salaminiou, 2005;Voriadaki, Simic, Espie, & Eisler, 2015), take on new perspectives (Duarte, 2012;Engman-Bredvik et al., 2016;Tantillo et al., 2015;Whitney, Currin, et al., 2012;Wiseman et al., 2019b), gain new skills (Duarte, 2012;Tantillo et al., 2015) and feel more empowerment (Engman-Bredvik et al., 2016;Salaminiou, 2005). Together, this helped people, particularly parents/caregivers, feel less guilty, scared, and anxious (Whitney, Currin, et al., 2012) and feel more confident (Whitney, Currin, et al., 2012;Wiseman et al., 2019b A common theme across several studies was that MFT led to a shift in the quality of family connection and dynamics (Baumas et al., 2021;Berit Støre Brinchmann & Krvavac, 2021;Duarte, 2012;Tantillo et al., 2015;Wiseman et al., 2019b). Commonly, parents/ caregivers felt MFT provided a new support network that helped people in all family roles feel less alone and isolated (Duarte, 2012;Engman-Bredvik et al., 2016;Salaminiou, 2005;Tantillo et al., 2015;Wiseman et al., 2019b), which was also echoed by some patients, albeit fewer (Duarte, 2012;Salaminiou, 2005). There was value placed on being able to observe and learn from other families who had similar experiences (Duarte, 2012;Whitney, Currin, et al., 2012;Wiseman et al., 2019b). In one study, participants struggled to differentiate the contribution that MFT made to their overall treatment compared to other elements of their treatment program (Baumas et al., 2021).
The challenging aspects of MFT were multifaceted. There were concerns by some about the potential for unhelpful comparisons to be made, the fact that individual family needs could not always be addressed, and that it was difficult at times to manage disparities regarding the different rates of recovery for each person in the group (Baumas et al., 2021). Of note, comparisons were also sometimes seen as helpful as they helped people feel validated and less isolated (Voriadaki et al., 2015).
A minority of parents/caregivers mentioned concerns that the group may set recovery backwards or that the patients may share unhelpful eating disorder "tricks" (Baumas et al., 2021;Salaminiou, 2005). The intensity of the group was also mentioned by some participants as both helpful and exhausting (Voriadaki et al., 2015;Whitney, Currin, et al., 2012;Wiseman et al., 2019b).
To understand changes in the patient experience during MFT, Voriadaki et al. (2015) collected data at different time points over four consecutive MFT days. They found that participants tended to move from anxiety and apprehension about attending, to noticing similarities and then feeling more settled. This helped people to become more aware of the illness and the role in it played in their relationships. By the end of the 4 days, the focus was shifted towards future coping and reflecting on progress. This matches data reported by Wiseman et al. (2019a) who specifically investigated the way family members perceived change to occur in MFT. No young people consented to participate in this study; however, parents perceived the treatment mechanisms to include the increased intensity, the experience of being with other families, family bonding, shifting of guilt and blame, improved parental confidence, and understanding the illness differently (Wiseman et al., 2019a).

| Meta-synthesis
Taken together, MFT is both perceived as helpful and leads to a wide range of improvements. It is also associated with several challenges for different participants at different time points. The benefits regarding eating disorder symptoms and other individual and family factors are reported by most, more so by family members than patients, and are reflected in robust quantitative and qualitative findings of eating disorder symptoms and physical health improvements by the end of treatment and often at follow-up.
As might be expected of any eating disorder treatment, MFT is also challenging, which is reflected in the more mixed family and caregiver quantitative and qualitative data. There are difficulties associated with the group process, such as concerns about the comparison that comes from being in a group, the intensity of MFT, and the realization of needing to try new things. However, these were balanced by a reduction in perceived isolation and support from the group. The low dropout rate further supports that the group process is engaging and acceptable.
The work required of participants during MFT coupled with the increased support afforded by coming together does yield rewards.
Eating disorder symptoms and physical health do consistently improve, but the difficulty and anxiety associated with reaching these changes are clearly reflected in the experiences, and more mixed quantitative findings, particularly regarding family factors. Nevertheless, while qualitative data suggest a key benefit of MFT is a reduction in isolation and an increase in solidarity, this may not generalize beyond the MFT group itself. Quantitative findings from one study indicate that self-report perceived social supports, isolation, and stigma more broadly do not significantly change during MFT.
One notable finding is that MFT can be described by the patient as more helpful for other family members, yet improvements are more consistently observed for the patient themselves. This fits with findings from one study, in which the majority of parents reported that the benefits of MFT for the patient came indirectly via themselves as parents (Salaminiou, 2005). This perceived disparity in helpfulness may also reflect the stage of recovery of participants at the time of data collection (typically during or at the end of treatment). What is perceived as unhelpful or challenging in the moment may be perceived as helpful and needed with hindsight. Salaminiou (2005)

| DISCUSSION
MFT has been used in clinical practice for decades (Gelin et al., 2018;Gelin et al., 2016) and practice guidelines often recommend including and supporting family members in eating disorder treatments (Hay et al., 2014;Hilbert et al., 2017;NICE, 2017). The current review highlights that MFT is associated with improvements in a range of individual and family factors. However, the evidence base is relatively small, with most studies underpowered and large heterogeneity in the MFT models tested.
Regarding the first aim of this study, it can be concluded that MFT is associated with physical and psychological improvement for people with eating disorders, across the age range, treatment setting, and diagnoses. This was almost unanimously reported across all studies, often with medium and large effect sizes. However, with the exception of a stand-alone 5-day model, all other MFT models were adjunctive treatments. As such, the unique benefits of MFT cannot be ascertained from the available data.
When MFT outcomes are compared to other types of treatment, the findings are mixed. There is evidence from two studies that 12 months of MFT alongside single-family therapy or treatment as usual may lead to better global outcomes and higher weight at followup for young people in an outpatient setting. This finding was reported in the largest study and only outpatient RCT that was identified in this review, meaning more weight could arguably be given to this finding. However, most other studies that included a comparison group found outcomes following MFT were equivalent to other treatments, both in outpatient and inpatient settings and across the age range. Furthermore, outcomes following intensive versions of family therapy have similar outcomes, regardless of whether it is provided in the single-or multi-family format. One potential benefit of MFT was its cost-effectiveness as an adjunctive adult inpatient treatment, although more data is needed here to form any firm conclusions.
Regarding the impact of MFT on individual and family/caregiver factors, the data highlights a range of benefits for all involved. For the patient, it is associated with improvements in symptoms of depression and anxiety, self-esteem, quality of life, and facets of emotion regulation capacity. However, MFT does not seem to be associated with improvements in perceived interpersonal problems or stigma.
Family members also report improvements in their own symptoms of depression, improvements in the negative aspects of caregiving, general well-being, and some changes in level of expressed emotion, although these data are mixed. This fits with evidence that not all aspects of expressed emotion may be as relevant for adolescents as they are with adults. A recent review found that parental emotional overinvolvement with adolescents was not associated with problematic symptoms or behaviors across a range of mental health diagnoses, and may even have some benefits (Rienecke, 2020). No changes were reported in the positive experiences of caregiving, level of caregiver anxiety, or perceived stigma and social supports. Data regarding changes in general family functioning are more mixed with some studies reporting improvements, others no change, and one a deterioration in functioning.
These findings match the data on the experience of MFT well.
These data show that MFT is generally valued, albeit with a fair degree of variability. Benefits are reported by most, especially the increase in perceived support by being in a group with people who have had similar experiences. The process can be exhausting and anxiety provoking, as might be expected of any psychological treatment, and may be part of what makes the treatment effective. Some participants raised concerns about the inevitable comparison that comes from being in a group and the realization of needing to try new things. However, available data on the process of change during MFT suggests these anxieties are alleviated quickly and by the third MFT day participants are more settled and perceived support from the group has increased. Given the experience of MFT can be intense and anxiety provoking, it is unsurprising that some aspects of caregiving and family functioning remain unchanged. Whether MFT experiences and learning lay a foundation for future benefits is yet to be determined and no data are available to report on this.
Despite the many benefits of MFT, the current review highlights some key areas for future research. Generally, the included studies were uncontrolled with small sample sizes, meaning most are likely underpowered. Even the RCTs tended to be small, with two of the three identified in this review having sample sizes of less than 50 participants. Furthermore, given most MFT models were adjunctive, the specific contribution and cost-effectiveness of adding MFT to other treatments remain unclear. Future studies are needed that examine the unique contribution made by MFT, both regarding outcomes and the experience of treatment, and whether the additional resources required of families and services are worth the benefit.

| Limitations
There are several important limitations to this review. First, only English language studies were reviewed, and the publication type was limited to peer-reviewed journal articles, book chapters, and dissertations (not conference abstracts). Furthermore, this is not an exhaustive review of MFT treatment models. Many theoretical and descriptive papers exist outlining MFT models that vary from those described here, without reporting any data or methodology for data collection.
Regarding the papers reviewed, the most notable are the small sample sizes and uncontrolled nature of study methodologies. Even though three RCTs were identified, two had sample sizes below 50.
Similarly, sample sizes for qualitative studies were also often small, with the voice of the person with the eating disorder limited or missing. Furthermore, there was a lack of diversity across the studies. The sample reported on was predominantly white and female with very little socioeconomic data reported. This makes interpretation of the data and conclusions from this review very tentative.
Finally, it is very hard to determine treatment response for people with bulimia nervosa and other eating disorder presentations. There was no data for adult MFT-BN and very limited data for young people.
MFT-BN findings are very preliminary.

| CONCLUSIONS
The current review suggests MFT is an effective treatment for anorexia nervosa and leads to improvements in individual, as well as some caregiver and family factors. The most robust finding is for young people seen in an outpatient setting. When added to singlefamily therapy, MFT may enhance outcomes compared to singlefamily therapy alone, although replication studies are needed. Several benefits afforded by MFT appear unique to the multi-family context; however, the impact of these benefits (e.g., increased support) cannot be determined from the current review. When compared to other types of treatment, MFT is generally non-inferior, although it is typically an adjunctive treatment, making it difficult to determine its value alone. Future studies are needed that specifically investigate the unique contribution and cost-effectiveness of MFT compared to other treatments. The evidence base also needs strengthening with higher quality studies with larger sample sizes and more diverse eating disorder presentations. Study designs that consider patient and family preferences, and previous treatment history are also needed, as well as clearer indication criteria for MFT and extended guidelines for the use of MFT in stepped care approaches.