14,096 studies were identified through database searching. Once duplicates were removed, the title and abstracts of 5,173 studies were screened, excluding 5,129 studies. Forty-four full-texts were then reviewed for eligibility, resulting in 34 articles being excluded. Reasons for exclusion were that studies were not focused on meal support as a first-line intervention (n = 20), were a conference abstract, poster, dissertation or other non-eligible paper type (n = 11), included the wrong patient population (n = 2), or not available in English (n = 1). A PRISMA flowchart of the study screening and selection process is presented in Fig. 1.

Quality assessment results are displayed in Supplementary Tables 1, 2, and 3. Purpose, background literature, study design and type, results’ statistical significance, analyses clinical importance and conclusions were described well for quantitative studies (n = 4) [15, 34,35,36]. Two studies did not include one or more of the following details: sample size justification, validity of outcome measures, description of the intervention, contamination and cointervention avoided, and drop-out reporting [35, 36]. Qualitative studies (n = 4) [37,38,39,40] mostly met the quality appraisal criteria except for two studies [39, 40] which did not report, or provided inadequate detail, for one or more of the following aspects: theoretical perspective, obtaining informed consent, identifying assumptions and biases of the researcher and reporting on the decision trail. Kells (2013) reported better outcomes in patients who received meal support compared to patients who did not receive meal support; however, it is unclear whether patient characteristics, severity of illness, length of diagnosis, and physical compromise were comparable to the intervention group. Mixed-methods studies (n = 2) [10, 41] tended to meet the quality appraisal criteria except for not integrating quantitative and qualitative results, and addressing inconsistencies in the results between these two components.

Quantitative (n = 4) [15, 34,35,36], qualitative (n = 4) [37,38,39,40], and mixed-methods design (n = 2) [10, 41] were identified. Specific methodologies included retrospective chart audits [15, 35, 36], a pre- and post- cohort study [34], semi-structured interviews [10, 39, 40], video analyses [37, 38], and surveys [10, 41].

Most studies (n = 7) were conducted within an inpatient specialist eating disorders unit [10, 15, 35,36,37,38, 40]. A private eating disorders clinic [34], child and adolescent public tertiary mental health community service [41], and home were also identified [34, 39]. The sample size ranged between 9 and 56 participants.

Study participants included patients diagnosed with an eating disorder (anorexia nervosa, bulimia nervosa, ARFID, binge eating disorder) (n = 5) [15, 34,35,36, 40] and in two, linked studies, a mixed group of patients with an eating disorder and some patients with a diagnosis of disordered eating not meeting diagnostic threshold [35, 36] receiving meal support; clinicians providing meal support (n = 4) [10, 37,39,39]; and parents and carers of patients with eating disorders (n = 1) [41].

As seen in Table 2, most studies (n = 7) examined patients with a primary diagnosis of anorexia nervosa [15, 35,36,37,38, 40, 41]. Patients with an eating disorder not otherwise specified (EDNOS) were investigated in two studies [36, 41]. Diagnoses of bulimia nervosa [41] and ARFID [34] were included in one study each. One study also included in their cohort some patients who did not meet the diagnostic threshold fo an eating disorder diagnosis [36]. Average Body Mass Index (BMI) ranged between 14 and 16.1 in the three studies that reported BMI [35, 36, 38].

Most (n = 6) studies included adolescents between 12 and 18 years old [15, 35, 36, 41], staff and patients at facilities that catered for adolescents aged 12 to 18 [37], or staff that treated adolescents 16 and over [38]. Two studies included patients under 12 years [34, 35]. Three studies included patients over the age of 18 [35, 36, 40]. One study included staff members for facilities treating patients of all ages [10]. Three studies involved exclusively [40] or predominantly [15, 35] female patients, and one study focused on mostly male patients [34]. Studies that did not specify gender focused on study characteristics and / or included staff members as participants.5)

Three studies, reported on clinicians of varying professional backgrounds delivering meal support: including nursing staff; clinical support staff; consultant psychiatrists; social workers, and; child welfare officers [37,38,39].

As seen in Table 3, staff to patient ratios varied across inpatient units starting from 1:1 [10, 34, 39] and ranging up to 1:10 [10]. Most (n = 9) studies provided meal support by a trained clinician [10, 15, 34,35,36,37,38,39,40].

Two studies examined the practical [38] or interpersonal [37] processes of meal support delivered within inpatient units. Practical processes consisted of three phases: preparatory (meals are served, and patients are asked to be seated at the table with their food), eating (patients and inpatient staff sit at the table and eat their meals, with support from the staff), and post-meal (patients finish eating and leave the dining room with the staff) [38]. Interpersonal aspects of meal support included: monitoring food intake, providing mealtime instruction, motivating and encouraging patients to complete the meal, expressing support and understanding, and providing psychoeducation [37].

The length of time of mealtimes varied from 30 min [34, 37, 38], to 60 min [10]. Supervised rest period immediately after the meal ranged from 15 to 60 min [10, 15, 38, 39]. While it is common practice in eatig disorder treatments (e.g., CBT) to use graduated exposure to ‘fear-foods’ in ARFID and AN, none of the articles described implementing a graduated approach to meal supervision.

Aesthetics of the dining room (e.g., size, shape, and setting of the dining table), timing of the meals to avoid delays, and consistency in approach were important aspects identified to alleviate distress in patients [10, 40]. Familiarity with the clinicians and having a standardised training approach were facilitators of meal support effectiveness [39]. Furthermore, clinicians providing empathic emotional support during mealtimes, such as encouraging patients to continue eating and reducing feelings of anxiety [10, 15], whilst also being assertive and firm around food consumption [10, 15, 37] was reported being effective. Distraction techniques utilised and identified as helpful included discussing unrelated topics, employing breathing techniques, playing games, and watching television or listening to radio [10, 15]. Staff, however, were uncertain around appropriate topics to discuss [10] and voiced that distraction could prolong eating time [