Interventions for autistic children with behavioural feeding needs: a scoping review Open Access

Approximately 40–90% of autistic children present with challenging eating and mealtime behaviour (Ledford and Gast, 2006) and are five times more likely to experience feeding difficulties than neurotypical peers (Sharp et al., 2017). The most common feeding behaviours that interfere with nutritional intake include selective eating, restricted diet, food refusal and challenging mealtime behaviours (e.g. prolonged mealtimes, challenges with sitting at the table for mealtimes, throwing/discarding food during mealtimes). Food selectivity and food refusal are the most frequently occurring challenges experienced by this cohort of children. Food selectivity involves eating a limited range of foods and refusing to try new ones (Bandini et al., 2010; Ledford and Gast, 2006; Ausderau and Juarez, 2013; Sharp et al., 2013; Gent et al., 2024). These challenging mealtime behaviours can be observed from an early age (Nadon et al., 2011). These behaviours can lead to both immediate and long-term medical issues, as well as impact developmental progress, physical growth and cognitive development (Volkert et al., 2016). However, research would indicate that whilst the mealtime and eating challenges impact significantly on the autistic child, they also have far-reaching consequences for the parents and family of the autistic child (Thullen and Bonsall, 2017; Ausderau and Juarez, 2013; Gent et al., 2024).

Evans and Rodger (2008, p. 98) describe mealtimes as “windows to family routines and rituals”. Routines, such as bedtimes, bathtimes and mealtimes, organise daily activities and structure family life (Schuck and Bucy, 1997). These routines help establish family rituals that express identity, like birthdays and Christmas celebrations (Evans and Rodger, 2008), which support family health and cohesion (Rodger and Umaibalan, 2011). The eating challenges of autistic children disrupt family mealtime routines, shifting focus from the family unit to the child’s needs, impacting family connectedness (Rodger and Umaibalan, 2011). Food refusal and mealtime behaviours increase parenting stress, affecting interactions and relationships (Marshall et al., 2015a, 2015b) and dinnertime is often considered the most stressful part of the day due to unpredictable behaviour and food preferences (Marquenie et al., 2011). Within this, interventions for mealtime challenges should consider their impact on the entire family. This scoping review aims to explore if these factors are considered during the assessment and intervention process within the relevant literature. The authors initially reviewed existing studies on paediatric feeding to gain an understanding of existing evidence-based interventions.

From 1999 to 2020, various systematic and scoping literature reviews on paediatric feeding disorders have been conducted. Some reviews cover general paediatric feeding needs (Kerwin, 1999; Sharp et al., 2010, 2017), while others focus specifically on autism (Ledford and Gast, 2006; Marshall et al., 2015a, 2015b). Certain reviews target younger age groups (0–6 years) (Howe and Wang, 2013; Marshall et al., 2015a, 2015b), while many address broader age ranges like 0–18 years (Kerwin, 1999; Sharp et al., 2010; Lukens and Silverman, 2014) and 0–25 years (Malzacher et al., 2020). Some are discipline-specific, focusing on interventions by Occupational Therapists for autistic children’s eating needs (Howe and Wang, 2013; Diaz and Cosbey, 2018). A review of these studies suggests that behaviourist-based interventions, which focus on systematically changing behaviours through various techniques and strategies, are the most documented and evidence-supported approaches for children with paediatric eating issues. Commonly used strategies include escape extinction (non-removal of spoon and physical guidance), differential reinforcement, and positive reinforcement. Many documented behaviourist interventions are intensive, conducted individually within clinical settings, and primarily clinician led. The primary outcome measures identified across these systematic reviews relate to oral intake of food (e.g. grams consumed, proportion of food taken, number of bites accepted, increase in variety and texture of food consumed, acceptance of food into the mouth, caloric intake), changes in anthropometric parameters (e.g. weight changes, body mass index changes), weaning from enteral feeding and addressing maladaptive mealtime behaviours. Few studies included in these reviews addressed the impact of eating challenges on parental emotional wellbeing. The studies completed by Sharp et al. (2013) and Greer et al. (2008) considered caregiver stress reduction. In addition, the review completed Howe and Wang (2013) highlighted that behavioural interventions effectively reduce caregiver stress. Although limited, some evidence-based interventions acknowledge the emotional effects of challenging mealtime behaviours on parents.

The authors of this scoping review sought to a. explore what interventions, other than those with a behaviourist underpinning, are available to children (0–12 years) with a primary diagnosis of autism who have feeding eating support needs and to map according to categories and key components, and establish the impact of the interventions, b. to determine the level of involvement of parents during the assessment and intervention process, and to see if measures were used that explore the impact of the eating challenges on the parent and the family unit, and c. to determine how the included studies report their interventions including timing, duration, dosage and the mode of delivery and using the TIDieR checklist (Hoffmann et al., 2014) to determine the quality in reporting the varying aspects of the interventions and identify any gaps that exist.

A scoping review was conducted to achieve these study aims. The Arksey and O’Malley (2005) framework guided the process. Additionally, the Preferred Reporting Items for Systematic Reviews and Meta-Analyses Extension for Scoping Reviews (PRISMA-ScR) checklist (Tricco et al., 2018) was utilised to ensure structured and comprehensive reporting of the review (Figure 2 in supplementary material).

Studies were included based on the following inclusion criteria as related to the nature of the study cohort and type of intervention, primarily:

• children with a primary diagnosis of autism and who present with feeding support needs;

• children with autism who may also have comorbid conditions (e.g. intellectual disability) are included as useful and relevant information may be lost if one was to focus on children with a sole diagnosis of autism given the significant number of children with autism who present with comorbidities;

• aged 0–12 years;

• any clinical intervention in any context (home, school, clinical, hospital) focusing primarily on behavioural feeding needs; and

• published in English only peer reviewed journals over the past 50 years (from 1973 to 2023).

Studies were excluded as follows:

• if participants did not have a primary diagnosis of autism;

• if the participants were aged over 12;

• if the intervention did not primarily target eating concerns;

• if interventions focused on dietary change for behavioural reasons (e.g. gluten and casein free diets/pharmacological);

• if interventions for feeding behaviours that were not disruptive to mealtimes or intake were described (e.g. pica, packing or rumination);

• if interventions, when reviewed by the authors, were considered to have a primary behaviourist basis; and

• if articles were not in a peer-reviewed journal. All grey literature including conference abstracts were also not included.

A search strategy was conducted in consultation with a subject librarian. Searches were conducted in Pubmed, Web of Science (core collection), PsycINFO, CINAHL (complete) and Embase. An initial search of PubMed, using controlled vocabulary terms, was conducted with the following four strands (“Autism Spectrum Disorder” AND “Eating” AND “Intervention” AND “children” (in title/abstract and with keywords) on 26th October 2022 which generated 905 articles. From further review of titles, abstracts, keywords and terms were analysed and lead to the revision of search terms. The final search strategies used “OR” and “AND” Boolean operators to combine controlled vocabulary terms with free-text terms (sample of search strategy in supplementary material).

The articles were imported into the reference management software EndNote™ (The EndNote Team, 2013). Reference lists of all articles selected for inclusion were screened for further sources.

The screening tool Covidence software was used for the screening process. The first (AMC) and second authors (KOD) independently used the screening tool to assess if studies obtained during the database searches were eligible for inclusion based on title and abstract information. The studies which met the eligibility criteria were then independently screened for inclusion based on a full text review. Any disagreements were resolved through discussion between the first and second authors with option for further discussion with third reviewer if required.

Some of the articles were imported as per Covidence software and most were imported by the first author from affiliated university library. Using a Covidence template, a data extraction template was developed to capture the characteristics of each included study, and the data was extracted from each study by the first and second authors independently.

The Template for Intervention Descriptions and Replication checklist (TIDieR) (Hoffmann et al., 2014) was used to structure and to evaluate reporting of the interventions in the included studies. This checklist was completed by two authors independently (AMC and KOD) and full consensus was reached on the findings (checklist available Figure 1 – supplementary material).

Nine studies (n = 9) were eligible for inclusion and were grouped into the following categories: Psychology-led interventions (33% – 3 studies), Occupational Therapy-led interventions (33% – 3 studies), Multi-disciplinary-led interventions – school based (22% – 2 studies) and Multi-disciplinary-led interventions – home based (11% – 1 study). Each study was allocated a reference number, and general characteristics were outlined in Table 1 (supplementary material). Key components of intervention are detailed in Table 2 (supplementary material) and parental involvement (and the nature of involvement) are included in Table 3 (supplementary material).

The results are discussed hereunder as per each aim:

1. Map the interventions available to this cohort of children according to categories and key components (Table 1 – supplementary material) and establish the impact of interventions across categories.

Three studies (Burton et al., 2021; Kuschner et al., 2017; Sharp et al., 2019) primarily led by psychologists included 51 participants with a mean age of 8 years, 95% of whom were male. Burton et al. (2021) combined Family-Based Therapy (FBT) and Unified Protocols for Transdiagnostic treatment (UP-C/A), referred to as FBT-UP, incorporating CBT and autism-specific strategies. Kuschner et al. (2017) and Sharp et al. (2019) developed and piloted the BUFFET program and MEAL plan, respectively. Kuschner et al. (2017) targeted slightly older children (8–12) with an intervention focusing on flexible thinking about food and informed decision-making.

Burton et al. (2021) primarily examined the successful achievement of goals identified by participants and their parents, including reduced usage (P1), the removal of a nasogastric tube (P2) and an increased range of foods consumed orally. Conversely, the other two studies (Kuschner et al., 2017; Sharp et al., 2019) involved piloting programmes with a primary focus on treatment satisfaction and acceptability.

Three studies (Hoyo and Kadlec, 2021; Lima de Oliveira and Souza, 2022; Miyajima et al., 2017), led by Occupational Therapists, which were conducted in the USA, South America and Japan between 2015 and 2022, were included in this category. Twenty-five children (80% male) with an average age of 4.46 years participated in these clinic-based interventions.

The studies by Hoyo and Kadlec (2021) and Lima de Oliveira and Souza (2022) highlight the therapeutic benefits of play. Hoyo and Kadlec (2021) emphasise using play to foster therapeutic relationships and support oral motor and sensory development. Lima de Oliveira and Souza (2022) explore play for food exploration and developmental purposes. Miyajima et al. (2017) focus on sensory integration therapy for improving sensorimotor development, which positively affects selective eating preferences. Two of these studies (Hoyo and Kadlec, 2021; Miyajima et al., 2017) suggested that understanding eating preferences in autism is crucial for guiding personalised interventions.

Miyajima et al. (2017) reported significant gains across all measured domains, including an increase in the variety of foods eaten and improved parental perception of diet balance. Hoyo and Kadlec (2021) highlighted functional gains like increased self-feeding independence and acceptance of puree when seated on a chair.

Ninety-six children (82% male) participated in both studies within this category (Chung et al., 2020; Yamane et al., 2020). The study by Chung et al. (2020) included an older cohort of children with a mean age of 11.5 years, whereas the study by Yamane et al. (2020) involved a younger cohort with a mean age of 4.5 years. The Chung et al. (2020) study was primarily conducted by school staff, although the authors had diverse professional backgrounds, including Nutrition, Nursing and Physiotherapy. The implementation of the Yamane et al. (2020) study involved various disciplines, including medical staff (Paediatrician, nursing staff), allied health professionals (Speech and Language Therapists, Occupational Therapists, Physical Therapists), a Nutritionist and other relevant pre-school staff such as cooking personnel and pre-school teachers.

Both studies focused on food exposure and reported gains in the variety and type of foods consumed. Chung et al. (2020) measured outcomes in week four by observing grams consumed at school and using parent questionnaires to assess habitual consumption. The study monitored children’s special diets over several years (1–3 years). Neither study addressed whether these gains were sustained over time or in different settings.

One study included in this category (Cosbey and Muldoon, 2017) examines the implementation and effectiveness of the Easing Anxiety Together with Understanding and Perseverance (EAT-UP^TM) program, which was developed by a Speech and Language Pathologist and Occupational Therapist. The programme employs various behaviourist strategies, such as behaviour momentum, reinforcement (positive/negative), prompting (visual, physical, verbal) and hierarchy of food acceptance. Additionally, it incorporates strategies from speech and language therapy (communication) and occupational therapy (environmental considerations), thus meeting the inclusion criteria for this review.

This paper highlights the positive impact of the EAT-UP^TM programme, showing increased food variety, acceptance of non-preferred foods and improved mealtime behaviours among the participants. The parent-led approach also empowered parents to manage problematic mealtime behaviours effectively.

• 2. Determine the level of involvement of parents during the assessment (Table 2 in supplementary material) and intervention and the consideration of the impact of eating needs on the emotional wellbeing of the parents and the family.

Parents were comprehensively included for assessment across all the examined studies, irrespective of their category. Many studies explicitly articulate the collaborative process involved in setting functional goals between parents and clinicians (Burton et al., 2021; Kuschner et al., 2017; Hoyo and Kadlec, 2021; Cosbey and Muldoon, 2017). From an intervention perspective, parental involvement was reported in 80% (9/11 studies) of the included research. The study by Chung et al. (2020) does not explicitly mention parental involvement, and the intervention detailed by de Oliveria and de Souza (2022) appears to be clinician-led, with the level of parental involvement not overtly specified.

The nature of parental involvement varied across the studies. Studies led by psychology and occupational therapy generally placed significant emphasis on parent training. This training encompassed direct instructional sessions with parents regarding the content and foundations of the programme (Kuschner et al., 2017; Sharp et al., 2019), the nature and aetiology of selective eating needs (Hoyo and Kadlec, 2021; Miyajima et al., 2017) and methods to facilitate parents to implement their learning through observation and coaching (Burton et al., 2021; Sharp et al., 2019) or role-playing scenarios (Lima de Oliveira and Souza, 2022). Notably, Miyajima et al. (2017) attributed some of the attrition from session one to session two to parental difficulty in implementing certain strategies learned. The content of several included studies lent themselves to meaningful parental involvement. For instance, Burton et al. (2021) emphasised parent–child dyadic interaction during feeding and the necessity for co-regulation, including the pacing of feeding, in later sessions. Kuschner et al. (2017), grounded in cognitive behavioural therapy (CBT), underscored the parental role within the cycle of anxiety and how parental reactions might inadvertently sustain this cycle, thus necessitating parental awareness. For Yamane et al. (2020), 20 parents were trained and given hands-on experience in preparing recipes used at the child development centre during lunchtime. In the study by Cosbey and Muldoon (2017), parents or caregivers acted as the primary interventionists, with the intervention taking the form of clinicians coaching parents on the implementation of selected strategies to address their child’s feeding needs. Direct teaching was provided as required.

As would be expected, most of the measures used across these studies focus on the child’s eating habits. Some attention is given to parental perspective on the eating behaviours (e.g. on the dietary imbalance) and on the parents’ perception of ability to actualise change in the feeding behaviour [e.g. Self-efficacy Assessment for Parents of Children with Selective Eating (SAPS) scale] as outlined by Miyajima et al. (2017). As several studies involved piloting programmes [e.g. Burton et al., 2021 (FBT-UP); Kuschner et al., 2017 (BUFFET); Sharp et al., 2019 (MEAL Plan) and Cosbey and Muldoon, 2017 (EAT-UP^TM)], feedback was obtained relating to parent’s perception of treatment and treatment satisfaction. Two studies incorporated baseline measures that referenced parents’ emotional well-being related to eating. Cosbey and Muldoon (2017) utilised the Family Quality of Life (FQOL) (Hu et al., 2011) and the Behavioural Paediatric Feeding Assessment Scale (BPFAS) (Crist and Napier-Phillips, 2001). Kuschner et al. (2017) examined the impact of the autistic child’s eating challenges on family daily functioning through parental report. While these measures were utilised at the baseline, they were not reflected in the reported outcomes. This suggests that limited focus was placed on exploring or addressing the emotional impact on the child, parents, or the family unit. Furthermore, no other study referenced a baseline or outcome measure related to this aspect of the eating challenge.

• 3. To determine how the included studies report their interventions including timing, duration, dosage and mode of delivery and b. using the TIdieR checklist (Hoffmann et al., 2014), to determine the quality in reporting the varying aspects of the interventions and determine any gaps that may exist within this (Figure 1 – supplementary material).

Overall, relevant information was included pertaining to the name, the rationale for the study, and the procedures involved in completing the study (for purposes of replication) in accordance with the TIdieR checklist. Sufficient detail regarding the timeframe, frequency, dosage, location and mode of the interventions was also provided. There was no consistent pattern across categories concerning the mode of delivery.

Several studies were conducted on an individual basis (Burton et al., 2021; Hoyo and Kadlec, 2021; Lima de Oliveira and Souza, 2022; Cosbey and Muldoon, 2017). A few psychology-led interventions incorporated parent-led groups and opportunities for parents to practice the learned strategies with their children (Kuschner et al., 2017; Sharp et al., 2019), while Miyajima et al. (2017) focused primarily on parent-led groups that also facilitated optional individual sessions with the parent and the clinician.

Sufficient information was provided regarding the location of the interventions, with all psychology and occupational therapy-led interventions being provided within a clinical setting. The multidisciplinary-led interventions were conducted both within a school setting (Chung et al., 2020; Yamane et al., 2020) and within the home (Cosbey and Muldoon, 2017). Although studies identified the location of the intervention, specific details regarding the infrastructure were not provided. For some studies, this lack of detail is not critical given the nature of the intervention (e.g. Kuschner et al., 2017; Sharp et al., 2019; Cosbey and Muldoon, 2017), which were primarily parent-led. However, for the study by Lima de Oliveira and Souza (2022), which has a strong sensory integration foundation, additional specifics concerning basic accommodation requirements would be beneficial given the differences in available resources from clinic to clinic.

Detailed information regarding the frequency and duration of sessions was provided, showing significant variation across and within intervention categories. The shortest intervention, a school-based program by Chung et al. (2020), lasted four weeks with sessions three times per week. The longest and most intensive intervention, led by Lima de Oliveira and Souza (2022), spanned 17 months with twice-weekly occupational therapy sessions, totalling approximately 136 sessions.

Most studies reported adequate details of the interventions in line with the TIDieR checklist with relevant information being reported relating to the initial 8 criteria. Nearly half of the studies (e.g. Hoyo and Kadlec, 2021; Cosbey and Muldoon, 2017; Burton et al., 2021; Sharp et al., 2019) detailed how interventions were tailored to the child’s specific needs (TIDieR items 9 and 10). Cosbey and Muldoon (2017) used assessments, parental interviews and baseline sessions to tailor the intervention to the individual child. Burton et al. (2021) aimed to increase oral intake and wean participants from nasogastric feeds. Conversely, Hoyo and Kadlec (2021) focused on transitioning a child to spoon feeding. These studies had small sample sizes (1–3 children), were delivered on an individual basis, allowing for bespoke approaches. The study conducted by Sharp et al. (2019) involved 38 participants. Interventions were customised based on parent-provided information and incorporated strategies suited to family routines and strengths, despite primarily being delivered through parent-led sessions. It is noteworthy that three of these studies that emphasised the importance of tailoring interventions to meet the individual needs of children also included follow-up evaluations: at 20 weeks into the programme (Sharp et al., 2019), nine months post-programme (Hoyo and Kadlec, 2021) and six months from the programme’s commencement (Cosbey and Muldoon, 2017). These studies demonstrated that, upon follow-up, the participating children generally retained the targeted improvements achieved during the intervention, such as an increased variety of foods consumed, reduced challenging behaviours, and enhanced parental actions to encourage better mealtime behaviours (Cosbey and Muldoon, 2017). Furthermore, one study provided information regarding treatment fidelity (Sharp et al., 2019).

Outcome of this scoping review would indicate that education/training of parents was integral to intervention across the majority of the nine studies. The nature of the parental training took many forms, depending on the content of the intervention, but typically it involved facilitating parents to have a greater understanding of the nature of the eating and feeding behaviours (Miyajima et al., 2017), upskilling parents on the specific strategies from within that specific theoretical framework (Cosbey and Muldoon, 2017; Kuschner et al., 2017) or facilitating the feeding relationship between the parent and the child (Hoyo and Kadlec, 2021). Since children mostly eat at home with parents as their main caregivers, parental involvement and providing parents with a better understanding of eating preferences appear to be essential for intervention. The authors of this paper conclude that understanding the rationale for the eating preferences and mealtime behaviours can allow for more realistic functional goal setting. Irrespective of the approach adopted, adaptation of the learned strategies is required to meet the individual unique child’s needs (Sharp et al., 2019; Burton et al., 2021). Indeed, this may be instrumental in ensuring that gains made during the “active” stages of intervention are sustained and generalised to all relevant contexts within the child’s life.

A myriad of other components was utilised across the studies including food exposure (Burton et al., 2021; Kuschner et al., 2017; Yamane et al., 2020; Cosbey and Muldoon, 2017) and environmental modification (Yamane et al, 2020; Cosbey and Muldoon, 2017). Although the primary focus was on interventions without a behaviourist foundation, it is noteworthy that many of the included studies incorporate strategies grounded in behaviourism, such as reinforcement and prompts (Burton et al., 2021; Yamane et al., 2020; Cosbey and Muldoon, 2017). Given this and considering the range of disciplines involved in intervention provision, and the fact that some strategies used in the studies were discipline-specific (e.g. sensory integration by Lima de Oliveira and Souza, 2022), it is important to recognise the need for multi-disciplinary interventions. Interventions to address the eating challenges should incorporate strategies from various professional backgrounds to address the unique and complex feeding needs of each child (Lukens and Silverman, 2014; Sharp et al., 2017).

The most noted baseline and outcome measures obtained related to the increase in type, number and/or category of food (Hoyo and Kadlec, 2021; Lima de Oliveira and Souza, 2022; Miyajima et al., 2017; Chung et al., 2020; Yamane et al., 2020; Cosbey and Muldoon, 2017). Some studies also identified improvement in mealtime behaviours as outcome measures e.g. participating in family mealtime (Chung et al., 2020; Sharp et al., 2019). Long-term gains could not be determined as only a few studies had a follow-up (Sharp et al., 2019; Hoyo and Kadlec, 2021; Cosbey and Muldoon, 2017). Within the studies included in this review, there was little reported consideration given to the emotional impact of the mealtime behaviours on parents and on the family unit. Given the far-reaching impact of the eating challenges, an intervention is required that emphasises a family-centred approach (Burton et al., 2021). This intervention needs to be informed by a detailed and holistic assessment of the child’s needs (Marshall et al., 2015a, 2015b), while also considering the impact of mealtime behaviours on the family. This will result in a more tailored approach that addresses both the child’s specific mealtime requirements (Sharp et al., 2019) and the familial and social implications of the needs.

This scoping review identifies several areas for future research. Future studies should aim to determine the elements of a comprehensive assessment that addresses eating challenges experienced by children and the impact on the parents, and family units. Reaching a consensus on the nature of assessments could help develop targeted interventions and streamline approaches for this group of children. Furthermore, standardising the terminology used to describe mealtime behaviours of autistic children might facilitate better research and intervention outcomes.

Given that eating and mealtimes are an integral part of family life, it is important to further investigate the effects of disrupted eating patterns on these mealtimes. Examining parents lived experiences of raising an autistic child with eating challenges would provide valuable insights and guide considerations for effective interventions. Similarly, exploring the individual’s own lived experience and journey would offer significant understanding into their personal perceptions and impacts, thereby allowing for more precise intervention strategies.

Eating challenges are common among autistic children and can affect the entire family’s wellbeing. This review examined non-behaviourist interventions for these children. The findings suggest that most interventions focus on the child, neglecting the family context. There is a need for family-centred interventions based on comprehensive assessments that consider the child within their family unit. This approach should address both the child’s eating challenges and their impact on the family.

The authors acknowledge the work of Ms Isolde Harpur, Subject Librarian of Clinical Speech and Language Studies, Trinity College Dublin.

The supplementary material for this article can be found online.