Search results

Following de-duplication, the titles and abstracts of 358 studies were screened, resulting in 41 for full-text appraisal (Fig. 1). At this stage, eight studies were excluded: two were unavailable, two were an ineligible article type (one case study, one conference abstract), and in four, accommodation was not explicitly measured. Three additional studies were identified by citation searching and from correspondence with colleagues. Therefore, a total of 36 studies were included in this systematic scoping review.

Study characteristics

This review included data from 32 quantitative studies and four qualitative studies published between 2009 and 2023. Included studies were published in the United Kingdom (n = 11), the United States of America (n = 6), Canada (n = 6), Italy (n = 4), Spain (n = 4), Australia (n = 2), Germany (n = 2) and Iceland (n = 1). Of the quantitative studies, there were various study designs, including 10 randomised trials, 13 longitudinal studies and nine cross-sectional studies. The study characteristics and their outcomes are summarised according to study design in Tables 1, 2, 3 and 4.

Measurement tools utilised across studies

The vast majority of the quantitative studies utilised the Accommodation and Enabling Scale for Eating Disorders (AESED; [51]) to measure accommodation. The AESED is a 33-item self-report measure that assesses the general tendency to engage in accommodation and enabling behaviours among parents and carers with EDs. The AESED has a five-factor structure (i.e., 1) Avoidance and Modifying Routine, (2) Reassurance Seeking, (3) Meal Ritual, (4) Control of Family and (5) Turning a Blind Eye), which have demonstrated adequate internal consistency, convergent validity, and sensitivity to change after family interventions. Higher scores on AESED subscales and total score indicate greater levels of family member accommodation and enabling behaviour.

Other tools to measure accommodation that were used in studies included the Family Coping Questionnaire for Eating Disorders (FCQ-ED; [12]) in one study [39], and the Family Accommodation Scale—Anxiety (FASA; [32] in one study [53].

Participant characteristics

A total of 1,839 people with EDs were included in the studies (N reported in 30 studies), who were mostly female (93%). The patient groups included across articles included AN (n = 1,517), EDNOS (n = 111), ARFID (n = 58), BN (n = 59), BED (n = 17) and OSFED (n = 2). The age of the people with EDs was reported by 22 studies and when pooled had a mean ± standard deviation (M ± SD) of 19.3 ± 7.1 years.

A total of 3,029 carers participated in the studies. The vast majority of carers were parents of people with EDs (n = 2,693), but also partners (n = 127), siblings (n = 16), grandparents (n = 5) and friends (n = 7). The pooled age of the carers was 49.2 ± 7.1 years, reported by 15 studies. Only 17 studies reported whether carers were living with the person with an ED, and amongst these studies 79% were cohabiting.

Study quality

The methodological quality of the articles was assessed with the JBI Critical Appraisal Tool. The studies were divided into four categories based on their design: randomised controlled studies (n = 10), longitudinal (n = 13), cross-sectional (n = 9) and qualitative studies (n = 4) (Tables S3-S6).

The quality assessment for randomised trials indicated that though true randomisation was used for all the studies, the nature of the interventions were such that it was impossible to blind the participants. The outcome assessors could have been blinded, which was done in six studies but not in three other studies. Most studies reported numbers completing follow-up or the reasons for attrition, and all studies measured outcomes in a reliable way, used appropriate statistical analysis and trial methodology. Overall the studies had low-to-moderate risk of bias; there were less than 50% negative or unclear responses on the quality assessment tool. The longitudinal studies overall had low risk of bias, although several studies didn’t identify and/or control for confounding factors (n = 7), or address incomplete follow-up (n = 5). Amongst the cross-sectional studies, the clear criteria to include in the sample were not included in three studies and the confounding factors were clearly identified and dealt with only in four studies, although outcomes were measured in a reliable way and appropriate statistical analysis was used in all studies. The methodology, representation of the participants’ voices and conclusions drawn were appropriate in all qualitative studies. However, there was lack of clarity around congruity between the stated philosophical perspective and the research methodology in two qualitative studies. The influence of the researcher on the research, and vice- versa, was not addressed in all the four studies.

Study resultsRandomised trials

The data from the 10 included randomised trials are tabulated in Table 1. The ten studies included information from people with AN (n = 707), BN (n = 38), BED (n = 6), ARFID (n = 4) and EDNOS (n = 21). A total of 1,193 carers participated across the trials. Four of these randomised trials used the intervention “Experienced Carers Helping Others” (ECHO) with three using treatment-as-usual (TAU) as comparison arm [21, 22, 33], and one compared standard ECHO to ECHO with coaching [16]. The other six studies used a diverse range of psychoeducation-based interventions [7, 18, 36, 44, 45, 50].

Table 1 Characteristics and results of randomised trials

Of the four ECHO trials, all found nominal reductions in AESED scores after ECHO, although this was not significantly different from TAU arms. In a study of mainly adult inpatients, there was no significant difference between arms at post-treatment, 6-months or 12-months [21]. A later publication based on follow-up data at 24-months from the same trial from Magill et al. [33] found nominally lower scores in the ECHO group compared to TAU. In an outpatient study of adolescents [22], there were nominally greater reductions at 6-months in the combined ECHO groups (ECHO and ECHO with guidance) in comparison to TAU. In the study comparing 6-weeks of ECHO and ECHO with telephone coaching, reductions in AESED scores were found in both groups, with no added benefit of telephone coaching [16]. This study found ECHO was most effective in people with high AESED scores, and improvements in AESED was moderated by adherence to the intervention. Additionally, changes in AESED scores were related to improvements in carers’ depression scores, and global ED functioning in their loved ones.

Three studies examined skills-based workshops for carers [36, 45, 50]. In a study comparing six 2-h collaborative care skills workshops to psychoeducational workshops over three months [45], there were reductions in AESED scores from baseline to follow-up in both groups, and reductions from baseline to post-intervention in the psychoeducational workshop group. There were no differences in AESED scores between groups at any time-point. In another comparison study of six 2-h skills-based workshops and psychoeducational workshops [50] both arms showed reduction in AESED scores at three-months and at a three-month follow-up, with no arm having superior reductions. Finally, in a study comparing two 2.5 h psychoeducational/communication skills-based workshops with waitlist participants, there was no greater reduction in AESED scores in either arm despite both showing a nominal reduction [36].

Two studies investigated web-based interventions [7, 18]. One study compared an eight-week web-based psychoeducational intervention for carers with a two-day psychoeducational workshop [7]. Both interventions were based on the same psychoeducational materials, with the web-based intervention occurring at the user’s own pace over eight weeks, and the workshops occurring over two 7-h workshops one month apart. From baseline to post-intervention, and from baseline to three-months, there was a reduction in AESED scores in both groups, which were not significantly different from one another. Another study compared a four-month “overcoming anorexia online” web-based intervention to four months of charity support [18], finding no difference between arms in AESED scores after treatment or at a six-month follow-up.

Additionally, one study compared a video-based intervention for carers (based on the interpersonal maintenance model) to TAU [44]. Both arms again reported a reduction in AESED scores at three-months, with no difference between arms.

Longitudinal studies

Study characteristics and findings of the thirteen longitudinal studies are included in Table 2. These studies collectively included information from patients with AN (n = 396), BN (n = 18), BED (n = 11) and EDNOS (n = 42), OSFED (n = 2) and ARFID (n = 54), and 721 carers. There was significant heterogeneity across the studies in terms of the type of intervention between the baseline and follow-up time-points. Four studies broadly included hospital programmes for young people that actively involved both the patients and their carers [5, 8, 20, 63], one of which was a study examining transitions to home treatment for young people with AN [20]. Seven studies included carer-focused interventions [15, 34, 38, 42, 43, 47, 53]. One study examined the use of a couples-based intervention for people with BED [66]. Finally, one other study examined pre and post outcomes in parents of children with AN who were hospitalized for medical stabilization [58].

Table 2 Characteristics and results of longitudinal studiesStudies of hospital programmes involving patients and carers

In three studies, changes were found in AESED scores of carers from baseline to post-intervention time-points [8, 20, 63]; this was not reported in one study [5]. In a study investigating home-based treatment for young people with AN [20] and a study investigating a partial hospitalization programme for young people with AN and ARFID including family involvement [63], changes in AESED scores were found from pre- to post-treatment. In another study investigating 25-session Family-Based Therapy for transition-aged youth, a reduction in AESED scores were observed from baseline to post-treatment, although this was not sustained at a 3-month follow-up [8].

Three studies examined the predictive value of AESED scores with differing findings [5, 8, 63]. In one study, higher AESED scores in parents predicted higher EDE-Q global scores in patients at discharge after participation in a comprehensive intervention using CBT-ED for patients and components aimed at reducing accommodation in carers [5]. However, in another study changes in AESED scores from pre- to post- intervention were not associated with changes in EDE-Q scores in patients or weight restoration [8]. Additionally, in the study by Wagner et al. [63], there was no association between changes in AESED scores of carers and treatment outcomes of patients.

Studies examining carer-focused interventions

One study utilizing a group-based “Therapeutic Conversation Intervention” rooted in four theoretical frameworks did not find reductions in AESED scores from baseline to post-intervention, although significant reductions were found from baseline to a 4-month follow-up time-point [15]. In another study examining the effects of a two-day emotion-focused family therapy workshop for carers of people with different EDs, significant decreases in AESED scores were found from baseline to a 6-month follow-up [42]. Another study utilizing weekly 2-h workshops aimed at developing collaborative care skills found significant reductions in total AESED scores from baseline to post-treatment and follow-up [43]. Reductions were found in all subscales other than two (“Meal rituals” and “Control of family”). One study found a reduction in AESED scores from baseline to a four-week follow-up in carers using an online virtual grocery store, which was utilized as an educational tool [34]. Finally, one study examining a 12-week parental intervention aiming to improve the food-related flexibility of children with ARFID via reducing accommodation in parents (SPACE-ARFID), reductions in FASA scores were observed from baseline to post-treatment [53].

Two studies [38, 47] performed secondary analyses from data from a RCT reported in the previous section [22]. One study performed a network analysis, yielding the paradoxical finding that baseline AESED scores in carers positively predicted the BMI of patients at a 1-year follow-up [38]. The other study used polynomial hierarchical linear models to examine the relationship between baseline AESED scores in parents and changes in AN symptoms in their offspring [47]. This study found that patient’s symptoms followed a quadratic curve, whereby they initially improved and then worsened. This interacted with the AESED scores of parents; where both parents had low AESED scores, the patient’s AN symptoms decreased in a quadratic manner, whereas where both parents had high AESED scores, their symptoms increased in a quadratic manner. Overall, the best outcomes were reported for patients whose parents were both low in accommodation, worse for patients whose parents were both high in accommodation, and moderate for patients where one of the parents was highly accommodating and the other was not.

Other longitudinal studies

A study by Weber et al. [66] examined a 22-session CBT-based couples intervention for people with BED, finding that AESED scores in partners were unrelated to binge eating severity in patients both before and after the intervention. The authors measured the emotional arousal in patients and their partners during conversation, finding that although AESED scores were overall unrelated to emotional arousal, partners who were highly accommodating were more reactive to emotional arousal of their patients at baseline. However after treatment, this association was not found.

One study aimed to investigate symptoms of post-traumatic stress in parents of children with AN hospitalized for medical stabilization [58]. This study found that at baseline, maternal but not paternal post-traumatic stress symptoms were positively associated with AESED scores.

Cross-sectional studies

The study characteristics and data from the nine included cross-sectional studies can be found in Table 3. These studies included information from people with AN (n = 385), BN (n = 3) and EDNOS (n = 45), and 1,077 carers.

Table 3 Characteristics and results of cross-sectional studiesOverall levels of accommodation in carers

Several studies reported comparisons between groups regarding the levels of accommodation in their caregiver samples. Two studies found that mothers and fathers of people with different EDs [3, 4] and AN specifically [39] had similar levels of accommodation, however in the latter study, mothers reported higher levels of avoidance. Additionally, accommodation was associated with shorter illness duration and families being “joined” [39]. In another study, accommodation scores were found to be higher in mothers than in fathers of young people with AN/EDNOS-R [46].

One study compared accommodation in carers (mostly parents) of people with AN and people with BN, finding greater levels of accommodation in carers of people with AN [54]. Additionally, another study that compared accommodation in parents of people with EDs to parents of people with substance-related disorders (SRDs) and healthy controls, found that both mothers and fathers of people with EDs reported higher accommodation than mothers and fathers of healthy controls [3, 4]. Additionally, fathers of people with EDs had lower accommodation scores than SRDs, but there was no difference between groups in mothers.

Accommodation and carer’s own mental health difficulties

In a study of parents of people with AN/EDNOS-R, maternal accommodation was found to be positively associated with their distress (measured by the Depression, Anxiety and Stress Scale; DASS) [46]. Similarly, a study based on the cognitive interpersonal maintenance model of eating disorders by Goddard et al. [17] highlighted that accommodation was positively correlated with carer distress and psychological control. However, structural equation modelling found that accommodation was not directly associated with carers’ distress but was indirectly linked via the carers’ own difficulties with eating.

Another study by Stefanini et al. [