Caregiver factors influencing family-based treatment for child and adolescent eating disorders: a systematic review and conceptual model

Introduction

Family-based treatment (FBT) is currently a leading evidence-based treatment for adolescents with eating disorders (American Psychiatric Association, 2023; National Institute for Health and Care Excellence, 2017). In FBT, parents or custodial caregivers are seen as the key factor in overcoming the eating disorder (Treasure et al., 2021). Treatment is delivered in three phases, which include caregivers assuming control of renourishing their child in phase one. In phase two, meal autonomy is gradually returned to the young person, followed by phase three which addresses any other developmental concerns or difficulties along with relapse prevention planning (Lock & Le Grange, 2013; Rienecke & Le Grange, 2022). An essential component of FBT is active parental participation and leadership, specifically in the restoration of the young person’s weight to a healthy level by serving, supervising and encouraging completion of all meals (Adams & Hovel, 2024; Treasure et al., 2021).

FBT has been validated for anorexia nervosa (AN) and bulimia nervosa (BN; Forsberg & Lock, 2015; Gorrell & Le Grange, 2021) and its use is being evaluated with other child and adolescent eating disorders (Lock, Sadeh-Sharvit & L’Insalata, 2019; Van Wye et al., 2023). However, remission rates for AN in adolescents are about 40%, with the need to improve outcomes highlighted (Lock et al., 2024). An essential component of FBT is parental participation and leadership, specifically in the restoration of the young person’s weight to a healthy level by supervising and encouraging completion of all meals (Adams & Hovel, 2024; Treasure et al., 2021). Motivation to activate the required level of caregiver participation is grounded in psychoeducation about the negative impacts of malnutrition, and the framing of carer support as essential to recovery (Lock & Le Grange, 2013; Rienecke & Le Grange, 2022). It is inherently assumed that parents will be capable of shouldering both the leadership expectations and emotional distress; however, it has been suggested that parents experience challenges that may influence their implementation of treatment strategies, including emotions, cognitions, behaviors and practical obstacles (Giles et al., 2022; Marchetti & Sawrikar, 2024; Wufong, Rhodes & Conti, 2019).

Previous studies have focused on quantitative outcomes in research settings, community health and qualitative outcomes from parental and clinician perspectives (Astrachan-Fletcher et al., 2018; Cripps, Serpell & Pugh, 2024; Datta et al., 2023; Dimitropoulos et al., 2024). However, a wider reaching review including all these settings has not been produced. A scoping review of FBT trials examined non-specific predictors, mediators and moderators of treatment, finding several caregiver-related factors were relevant to outcome within the results including parental self-efficacy, expressed emotion, parental behaviors, treatment alliance and family conflict (Gorrell et al., 2022). Caregiver-level psychosocial factors such as blame, guilt or lack of empowerment, as well as accommodation of the eating disorder, family dysfunction, caregiver mental health or practical barriers have also been cited as negative influences on treatment delivery or outcome (Arikian et al., 2007; Keel & Haedt, 2008; Lavender, 2020; Loeb et al., 2012; Shimshoni, Omer & Lebowitz, 2021; Tasaka et al., 2017). The influence of cultural factors on family eating disorder treatment have been explored, such as traditional parental division of labor, culturally-sanctioned eating behaviors, collectivist or individualist values, with recommendations to adapt FBT or utilize cultural beliefs as a therapeutic resource (Chu, 2022; Dimitropoulos et al., 2024; Ram & Shelke, 2023; Williams, Wood & Plath, 2020). Parental strengths may include protective factors which may assist parents or caregivers in implementing FBT. Secure parent–child attachment, supportive family members, a willingness to learn about FBT, socio-economic stability or the ability to adjust commitments to support treatment have all been explored as potentially helpful (Basker et al., 2013; Jewell et al., 2021; Rienecke & Le Grange, 2022; Wallis et al., 2017b). These factors have thus far been assessed independently and have not yet been unified into a framework for evaluating strengths and challenges.

Several iterations of family treatment for eating disorders exist which emphasize the parental leadership role. In parent-focused treatment (PFT) the components of FBT are delivered separately with the clinician seeing the adolescent first, immediately followed by the caregiver session (Le Grange et al., 2016). PFT has been identified as potentially more effective than conjoined FBT in the presence of parental high expressed emotion (Allan et al., 2018; Eisler et al., 2000; Hughes et al., 2019). A number of intensive full-day programs incorporate FBT principles including intensive individual family therapy (IFT) delivered in approximately 40 h across one week by a multidisciplinary team, and multi-family therapy (MFT) in a group formats with five to eight families receiving four to five full days of treatment together across one week (Marzola et al., 2015; Rockwell et al., 2011). A longer-term day treatment program (DTP) is delivered in groups of five to eight families with 30 h of multidisciplinary treatment delivered weekly across approximately three months (Krishnamoorthy, Shin & Rees, 2023). It has also been argued that some families may not respond to individual FBT, and that the reduction in isolation and community created by a group setting in MFT as well as additional therapeutic activities may help support these families (Bruett et al., 2022; Dawson et al., 2018; Dennhag, Henje & Nilsson, 2021; Rienecke, 2017). MFT has also shown promising results when delivered as an adjunct to single-family FBT (Funderud et al., 2023). DTP has been shown to be effective and also support the FBT core tenet of parental involvement while supporting adolescents who require a higher level of care (Martin-Wagar, Holmes & Bhatnagar, 2019). Some differences exist between PFT, FBT and family therapy for anorexia nervosa (FT-AN); namely, PFT and FT-AN do not usually include a therapeutic family meal, and the four phases of FT-AN are condensed to three phases of FBT (Gorrell, Simic & LeGrange, 2023). However, the responsibility of the parent for nutritional rehabilitation is a key component of all derivations (Gorrell & Le Grange, 2021; Rienecke, 2017). While these versions of family treatment have demonstrated efficacy (Forsberg et al., 2023; Madden, Hay & Touyz, 2015), no protocol exists to assess caregiver factors, which could identify which format may be best suited to a particular family.

Methods

The research question was devised using the population, intervention, comparison and outcome (PICo) framework (Stern, Jordan & McArthur, 2014), and registered with the PROSPERO database of systematic reviews, ID number CRD42022338843. Preferred reporting items for systematic reviews and meta-analyses (PRISMA) guidelines were utilized for protocol development (Moher et al., 2009), with the research software Covidence used to screen database search results for duplication and relevance (Veritas Health Innovation, 2022). The nominated population included parents of children or adolescents with a diagnosed eating disorder listed in the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition, Text Revision (DSM-5-TR; American Psychiatric Association, 2022) for which FBT has been used. Included were: anorexia nervosa (AN), atypical anorexia nervosa (AAN), bulimia nervosa (BN), avoidant restrictive food-intake disorder (ARFID), binge-eating in the context of bulimia, and other specified or unspecified eating disorders (Forsberg & Lock, 2015; Gorrell et al., 2022; Lock, Sadeh-Sharvit & L’Insalata, 2019; Rienecke & Le Grange, 2022).

A Boolean search string was created by the research team (AM, KR, AR) and performed by AM/KR with terms derived from a scoping search of the literature using the terms eating disorders, parents and adolescents. A university librarian and a third-party research assistant reviewed the keywords and search strategy to address the risk of bias, and refine the search string. Keywords are listed by category in Table 1. Keyword categories included terms related to the involved adult/s, the intervention, the disorder, the client, and influencing factors. Databases searched included EBSCO Host, Embase, ProQuest, PubMed Central, SCOPUS and Web of Science. Searches were limited to articles published after and including the seminal literature on FBT published by Russell et al. (1987) until July 6, 2024. Results were restricted to peer-reviewed journal articles, theses and dissertations. Keywords were separated internally by the Boolean operator “OR” and between categories by the “AND” operator, with the asterisk “*” used as a wildcard to capture alternative spellings.

Inclusion criteria reported in PROSPERO were (1) a client population of families with at least one adult caregiver of children or adolescents aged 6–18 with an eating disorder as defined by the DSM-5 TR (American Psychiatric Association, 2022). (2) The families were required to be engaged in a version of FBT including traditional conjoined FBT or FT-AN, PFT or FBT-informed treatments including DTP, IFT or MFT. (3) An assessment or interview of parents or caregivers must have been conducted. Exclusion criteria included (1) clients over age 18 or below age 6, (2) clients who did not live with adult caregivers or (3) where the caregivers were not assessed or interviewed, (4) the lack of an eating disorder diagnosis or diagnoses for which FBT is not typically used (e.g., pica or rumination), or (5) clients receiving another form of eating disorder treatment, such as systemic family therapy, acceptance and commitment therapy, cognitive behavioral therapy or dialectical behavior therapy, without adjunct FBT.

Preliminary searches returned a wide range of study methodologies, including randomized controlled trials (RCTs), quantitative and qualitative studies, studies reporting FBT clinician observations, and clinical case studies. While a wide variety of articles was intentionally sought in order to capture as many known influential factors as possible, it was identified that such broad results could limit the ability to draw conclusions. The decision was made to report results in an evidence hierarchy by study type according to health research standards (American Psychological Association, 2006; LoBiondo-Wood & Haber, 2022). Three study methodologies appeared in the search results which are classified as tier six, including qualitative studies with parents and with FBT clinicians, and clinical case studies. Consensus on inclusion of these methodologies was achieved based on the arguments that synthesis of qualitative and quantitative research is essential in health care research and should include all articles relevant to a research question (Dixon-Woods, Fitzpatrick & Roberts, 2001). It has been argued that eating disorder research on factors influencing outcome should include FBT clinician perspectives as they hold the responsibility for clinical decision making (Dimitropoulos et al., 2024). Consensus on the inclusion of clinical case studies was reached in accordance with the argument that psychological evidence-based practice should consider case study findings when change to assessment or process is proposed (Willemsen, 2022). In order to narrow the results, consensus was achieved among the authors to focus on eating disorders for which FBT has been validated in RCTs, resulting in the removal of studies focused on ARFID and binge-eating disorder.

From these revised criteria, 1,994 articles were returned including 1,970 from databases and 24 from citation searches. Citations were exported to EndNote and uploaded to Covidence. Duplicates were removed automatically with 691 studies screened by title and abstract against inclusion criteria by two independent parties; the lead investigator (AM) and a research assistant. Disagreements were resolved by the supervisory investigators (KR, AR), with 164 articles selected for full-text review. Initial quality assessment of articles was conducted by AM using the Joanna Briggs Institute critical appraisal tools for health care research specific to each study design (Lockwood, Munn & Porritt, 2015). A subsequent second round of quality assessment was completed by AM and replicated by an independent research assistant to ensure included articles had sufficient quality, academic rigor and a low risk of bias, using the McGill University mixed methods appraisal tool (MMAT; Hong et al., 2018) as studies were both qualitative and quantitative. Discrepancy resolution and confidence in the body of evidence were achieved through consultation with KR. This process resulted in 39 included studies which met MMAT quality and risk of bias standards, represented in Fig. 1. Studies were conducted in Australia (7), Canada (5), Europe (2), the United Kingdom (7) and the United States (18). The results of 3,782 measures or interviews were reported, completed by parents or caregivers to adolescent participants who were 86.9% female, with a mean age of 14.87 (0.74). Sixteen studies reported ethnicity, of which 77.95% of families were Caucasian. Twenty-nine studies reported family structure, of which 68.41% were intact families. Four studies reported observations of caregiver barriers from interviews with 117 FBT clinicians.

Heterogeneity of the data including varying study design, description of caregiver participants and analysis and variables measured excluded the possibility of meta-analysis; therefore, a systematic-narrative hybrid approach (Siddaway, Wood & Hedges, 2019; Turnbull, Chugh & Luck, 2023) was used and synthesis without meta-analysis (SWiM) guidelines were followed (Campbell et al., 2020). The Cochrane data extraction template (2023) was adapted to collect and organize the results by AM with oversight by KR/AR. Methodologies included 24 quantitative, 12 qualitative and three case studies. Twenty-four studies were classified tier 1, 2 or 3 including systematic reviews of RCTs, RCTs and quantitative studies, and fifteen studies were classified tier 6 including twelve qualitative studies and three clinical case studies. An overview and summary of included studies is organized by tier and first author, and presented in Table 2.

Results

Synthesis of the data was achieved by grouping the findings as recommended by Campbell et al. (2020) into thematic domains. The initial factor groups were those identified during formulation of the research question: emotions, cognitions, behaviors or practical barriers. Within these groups, eight domains were identified following the systematic-narrative hybrid process (Siddaway, Wood & Hedges, 2019; Turnbull, Chugh & Luck, 2023). Specifically, each included article was organized by the independent variable or phenomenon of interest measured. As data was extracted, reoccurring variables measured within studies informed the generation of eight domains by AM with oversight from KR/AR. They are defined in the following section as: capacity, confidence, readiness, caregiver externalizing factors, caregiver internalizing factors, caregiver food-related factors, family function, and support networks. These domains were subsequently organized into a conceptual model using content analysis and frequency of appearance of each variable. It is represented visually in Fig. 2 at the conclusion of this section.

The review process highlighted that self-efficacy and expressed emotion, both well-known concepts or categories of behavior (Bandura, 1977; Brown, Birley & Wing, 1972), have been updated in the context of eating disorder research (Cherry et al., 2018; Rhodes et al., 2005). The definition of these concepts as well as methods of measurement varied within the results. This review attempted to synthesize the results by examining the individual qualities or behaviors attributed to expressed emotion and to self-efficacy, and reviewing the ways in which they influence eating disorder treatment.

Discussion

This systematic review included 39 studies which examined caregiver factors influential to FBT outcome including emotions, cognitions, behaviors, practical barriers and other factors. The recurring variables were measured by frequency of appearance, which resulted in the generation of eight domains including capacity, confidence, readiness, caregiver externalizing factors, caregiver internalizing factors, caregiver food-related factors, family function, and support networks. These domains and their constructs were organized into a visual conceptual model of caregiver factors influencing treatment (Care-FIT).

The review had the unanticipated result of highlighting the complexity of some well-known concepts within the field; that is, self-efficacy and expressed emotion. In the context of eating disorder treatment, Rhodes et al. (2005) defined self-efficacy as the ability of carers to adopt a primary role in managing the young person’s eating disorder treatment. The results of this review identified caregiver self-efficacy as predictive of successful FBT delivery and weight restoration (Byrne et al., 2015; Robinson et al., 2013). It was initially planned to include self-efficacy as a separate domain given the appearance of the concept in 18 of 39 articles; however, definition of the concept varied widely within the review results, with seven of the 18 articles not providing a definition. Eleven articles related self-efficacy to confidence or empowerment, or both, while four articles related self-efficacy to either ability, capacity, competence or resources. The present review explored these subcomponents individually and categorized them according to frequency of appearance. It is proposed that ability and competence are related to a caregiver’s capacity to carry out their part in the treatment, and that resourcing, caregiver health and caregiver burden affect one’s capacity to engage. The term carer empowerment was used interchangeably with confidence, and consistent with previous findings (Hagan et al., 2023; Robinson et al., 2013) it is suggested that both of these components can increase self-efficacy.

Criticism, hostility, emotional overinvolvement, warmth and positive remarks, were identified in this review as components of external expressed emotion. Bohon et al.’s (2023) findings grouped criticism, hostility and emotional overinvolvement together and suggested that the presence of any of these factors might negatively influence treatment. However, other articles in the results examined expressed emotion components individually. For clarity and to avoid the confusion noted by Barrowclough & Hooley (2003) surrounding whether these variables should be categorized as behaviors or emotions, the outward expression of each component was examined within caregiver externalizing factors.

Several caregiver externalizing factors identified in this review could potentially influence outcome positively or negatively. Le Grange et al. (2011) found that warmth and positive comments predicted weight gain. Further, the authors hypothesized that the absence of warmth might negatively influence outcome (Le Grange et al., 2011), which is in line with findings in this review that maternal criticism was related to poor outcome (Allan et al., 2018), and less criticism predicted better outcome (Darcy et al., 2013; Hughes et al., 2019). Likewise, factors contributing to caregiver burden could be positively or negatively related to outcome, in particular the carer’s perception of burden level, with high or increased burden negatively influencing treatment and feelings of empowerment, and lower or reduced burden leading to better outcomes as well as increased self-efficacy (Baumas et al., 2021; Couturier et al., 2013; Dennhag, Henje & Nilsson, 2021; Dimitropoulos et al., 2017; Matthews et al., 2023; Matthews et al., 2018; Stewart et al., 2021; Thibault et al., 2023). Parental control over their child’s activities appeared to potentially influence treatment positively or negatively, with low control predicting dropout and higher control related to remission (Ellison et al., 2012; Wallis et al., 2017a). It is suggested that if carer behavioral factors can be identified and managed, treatment success rates may improve, aligning with the findings of Lock et al. (2006).

Within internalizing factors, findings related to the influence of parental certainty on outcome appeared complementary. The finding that intolerance of uncertainty reduced confidence and increased parental accommodation of the eating disorder (Konstantellou et al., 2022) aligns with the finding that excessive parental certainty about their child’s mental state lead to poor outcome (Jewell et al., 2021). This systematic review results highlighted that guilt appears to affect treatment differently based on when it is experienced. Although the FBT model adopts an agnostic etiology of the disorder and attempts to minimize parental guilt (Rienecke & Le Grange, 2022), one included study found that higher pre-treatment maternal guilt resulted in a better outcome (Dennhag, Henje & Nilsson, 2021), while self-blame and feelings of failure experienced within treatment if weight gain did not progress, led to reduced empowerment and poorer outcome (Wufong, Rhodes & Conti, 2019). These findings support earlier conclusions that guilt arising if the young person did not achieve weight restoration in FBT phase one led to a poor outcome (Conti et al., 2017; Lavender, 2020), and perhaps suggests that clinicians should assess for pre-treatment guilt as well as monitoring parental emotion through treatment and offering additional support if weight restoration is challenging. The influence of parental mental health in general could potentially have a two-way impact, given that poor carer mental health increases burden and therefore poor outcome (Couturier et al., 2013; Dimitropoulos et al., 2017; Forsberg et al., 2017), while improvements in carer mental health and reduced burden were related to treatment completion and good outcome (Dennhag, Henje & Nilsson, 2021; Stewart et al., 2021). Finally, the finding that caregivers’ subjective perception of their child’s eating disorder treatment was predictive of outcome suggests that clinicians should be attuned to caregiver distress and offer additional support to minimize the risk of disengagement (Matthews et al., 2018).

Some elements of family function historically thought to impact treatment were found to be less conclusive in this systematic review. The finding that socioeconomic status or intact families do not predict remission (Datta et al., 2023) may help address historical concerns raised by clinicians that FBT is not appropriate for some families, and supports previous arguments that FBT can be successfully delivered to these families (Astrachan-Fletcher et al., 2018; Loeb & Le Grange, 2009). However, the reports of high burden, isolation and need for longer treatment among single parents (Doyle et al., 2009; Lock et al., 2005) should not be ignored in light of Datta et al.’s (2023) findings, particularly when considered with Lock et al.’s (2006) finding that longer treatment length was related to dropout. Attachment insecurity is known to negatively impact eating disorder treatment in adults and with different therapy modalities (Rossi et al., 2022); however, the results of this review linked attachment to outcome indirectly through therapeutic alliance (Jewell et al., 2021). This phenomenon was addressed by Wallis et al. (2017b), suggesting that the level of parent–child relational containment during FBT phase one nutritional rehabilitation may have a reparative effect on attachment, making an insecure or anxious attachment at commencement less of a risk to treatment outcome.

Intensive FBT-based programs studied in the present systematic review included DTP, IFT and MFT. These programs do not include any less carer involvement than traditional outpatient FBT; however, significantly more clinical time is spent with the young person and with the family in groups, which had the effect of increasing caregiver support (Girz et al., 2013). The finding that group cohesion in MFT was a mechanism of change (Baumas et al., 2021) supports a suggestion by parents that individual FBT may not be enough, particularly when caregivers or families are isolated. It may be helpful to expand the use of MFT as an adjunct to FBT to support increased remission rates, in line with Funderud et al.’s (2023) study that found significant improvement when families received concurrent MFT and FBT.

Limitations

The 39 articles included in this systematic review originated from the United States, Canada, Australia, the United Kingdom and Europe, which limits their generalizability to non-English speaking or non-western countries. While research returned in the database search from other countries (e.g., India, China and Japan), did not meet inclusion criteria for this review, they appear to align with the present systematic review’s findings that family function, cultural influences, and carer emotion and behavior (including acceptance of parental responsibility for renourishment) were all important factors to adolescent eating disorder outcome (Basker et al., 2013; Chu, 2022; Ram & Shelke, 2023; Sysko & Hildebrandt, 2011; Tasaka et al., 2017). Also, some included studies reported limitations that their participant samples consisted of treatment completers or that dropout factors were not analyzed (Forsberg et al., 2017; Forsberg et al., 2014; Le Grange et al., 2011), and thus there is a risk that not all treatment dropout factors were captured. Byrne et al. (2015) averaged caregiver Parents versus Anorexia scores rather than reporting results individually due to inconsistent parental attendance, which limits the ability to draw conclusions about differences between mothers and fathers or parental consistency within those families. Bohon et al. (2023) used a composite expressed emotion score combining criticism, hostility and emotional overinvolvement. Further, some studies (Allan et al., 2018; Hughes et al., 2018) noted the lower participation rates of fathers, which they suggested could have affected statistical power. Differences in healthcare systems between the United States and countries with universal or single-payer systems may account for differences in treatment completion. Access to treatment resulting from the healthcare system was cited as a barrier by Binkley & Koslofsky (2017).

Conclusion

Considerable evidence supports the RCT efficacy and real-world effectiveness of FBT and its permutations on child and adolescent eating disorder treatment, with guidelines recommending its use (Le Grange et al., 2021; Simic et al., 2022). While studies have explored individual factors which may increase or reduce effectiveness of FBT (Rienecke, 2017), this present study is the first to review, consolidate and organize these factors into a single conceptual model.

The present systematic review identified several caregiver factors which appear to positively or negatively influence the outcome of adolescent eating disorder treatment. These factors were grouped into a conceptual model including eight domains: caregiver capacity, confidence, readiness, internalizing factors, externalizing factors, food related issues, family function and the carer’s support networks.

It is hoped that this conceptual model will assist researchers in future experimental design to further explore the degree to which these factors influence treatment and whether some factors are bidirectional, or whether multiple factors are inter-related or have an exponential effect. The model may assist eating disorder clinicians with case formulation and treatment recommendations. Clinicians might use the model to consider whether adjuncts to FBT are appropriate, or whether a particular FBT format or version might be more suited to a family. Overall, it is hoped that assessment of caregiver factors at commencement of treatment could assist clinicians in recognizing possible treatment obstacles to be addressed and carer strengths which could support client recovery.

Supplemental Information