Objectives
The objective was to update three published systematic reviews of interventions to improve the EDSD of children with neurodisability that were published before our research began.
This chapter outlines the methods and findings of the update of two of these reviews.5,6 The third review4 was updated by the original review team during the period of our study. The update review followed the original review protocol. The authors of that review provided pre-publication data and these have been included at the end of this chapter for completeness.
Background
Three recently published systematic reviews relevant to interventions for young children with neurodisability and EDSD were identified at the proposal development stage.4–6
Marshall et al.5 reviewed evidence for interventions to improve feeding difficulties for children aged < 6 years with ASD that were published between January 2000 and October 2013. The review included 23 studies, which were a mix of single-case experimental design (n = 12) and small-group studies (n = 11), all with five or fewer participants who were aged 2–5 years. A total of 46 children participated, 37 of whom had ASD. The interventions were based on operant conditioning to ‘increase desirable eating behaviours or decrease undesirable eating behaviours’.5 The risk of bias of each study was assessed using a tool developed for single-case experimental design by Horner et al.18 The primary outcomes of interest were increased volume of food and variety of intake, which were considered to increase desirable mealtime behaviour. The secondary outcome of interest was a reduction of inappropriate mealtime behaviours. The authors concluded that there was a low level of evidence to support these types of interventions for children with ASD and that ‘favourable intervention outcomes were observed in terms of increasing volume, but not necessarily variety of foods consumed in young children with ASD and feeding difficulties’.5
The NICE cerebral palsy guidance6 reviewed evidence generated by RCTs and observational studies of interventions for ‘management of eating, drinking and swallowing difficulties’ of children and young people with cerebral palsy aged < 25 years published before October 2014. The outcomes of interest in the review were:
physiological function of the oropharyngeal mechanism (as determined by clinical evaluation, videofluoroscopic swallow studies or fibreoptic endoscopic evaluation of swallowing)
changes in diet consistency that a child is able to consume (developmentally appropriate oral diet; texture and/or consistency of foods and fluids must be modified; supplementary feeding required)
respiratory health – presence of a history of confirmed aspiration pneumonia or recurrent chest infection (with or without pneumonia with suspected prandial aspiration aetiology)
nutritional status and/or changes in growth (weight and height percentiles)
child and young person’s level of participation in mealtime routine/length of mealtimes (time taken to feed)
psychological well-being of parents and/or carers
acceptability of the programme
survival.
The included publications comprised four RCTs and four cohort studies, which included a total of 235 participants who were aged 12 months to 21 years. The four RCTs all compared Oral sensorimotor therapy with routine therapy; each cohort evaluated a different intervention. One cohort considered the Innsbruck Sensorimotor Actuator and Regulator (ISMAR) compared with no ISMAR; one an Oral sensorimotor treatment; one a training programme delivered to children and caregivers; and one evaluated individual interventions delivered in combination (a multicomponent intervention), including carer training, behavioural interventions and Beckman Oral Motor Exercises.19 The risk of bias was assessed using the Cochrane risk-of-bias tool.20 Conclusions were not drawn from this review alone, but as part of the total guideline development process. This took account of stakeholder input to the evidence and consideration of the evidence by a committee made up of practitioners, professionals, care providers, commissioners, those who use services and family members or carers. Based solely on the published studies (n = 8) the Grading of Recommendations and Assessment, Development and Evaluation (GRADE) assessment rated the evidence to support these types of interventions in this population as being of very low to low quality.6
Morgan et al.4 considered interventions for children with any ‘neurologically based oropharyngeal dysphagia’.4 The outcomes of interest were amount/variety of food and eating behaviours. Their review included two papers assessing sensorimotor treatment in 55 cerebral palsy patients aged 4–21 years; the studies were small and were rated as being at high risk of bias, as assessed by the Cochrane risk-of-bias tool.20 Both of the papers were also included in the NICE6 review. Morgan et al.4 concluded that there was ‘insufficient high-quality evidence from RCTs or quasi-RCTs to provide conclusive results about the effectiveness of any particular type of oral motor therapy for children with neurological impairment’.4
Methods
We updated the review by Marshall et al.5 and the NICE cerebral palsy review6 of EDSD interventions using the methods of the original reviews. Marshall et al.5 provided the search strategy for the ASD review and the NICE review searches were based on the published search strategies. Updated searches were limited to 1 year before the date of the last searches undertaken for the original review, which allowed for delays in database updates. Protocols for the systematic reviews were registered on PROSPERO (www.crd.york.ac.uk/prospero/; accessed 11 November 2019). Full details of the search strategies are presented in Appendix 2.
Inclusion criteria
Studies were included in the update of Marshall et al.5 if they met the following criteria:
Population – children aged 0–8 years with a diagnosis of ASD. (The original review included children aged 0–6 years, but we extended the age range to fit with our definition of young children with neurodisability.)
Intervention – non-pharmaceutical behavioural or environmental interventions.
Outcomes – amount of food and/or variety of foods consumed and/or desired or undesired eating behaviours.
Study design – an experimental design was used to investigate treatment outcomes, including the use of a control group, within-group designs or single-case-based experimental designs replicated across at least four participants (to give some indication of repeated effects of interventions).
18Language – studies were published in English in peer-reviewed journals.
The update of the NICE review6 included studies that met the following criteria:
Population – children and young people aged 0–8 years with a diagnosis of cerebral palsy. (The original review considered young people aged < 25 years.)
Intervention – interventions that aimed to improve sucking/biting/chewing/swallowing of food, intake of food (amount and/or variety of food) and/or eating behaviours.
Outcomes – Co-ordination of chewing and swallowing, Increased amount and variety of intake, Duration of mealtimes, Nutritional status, Growth and Physical health (e.g. chest infections and mortality).
Study design – group experimental design.
Language – studies were published in English in peer-reviewed journals.
Screening/data extraction/quality assessment
For both review updates, two researchers (HT and LP) independently screened titles and abstracts to identify studies meeting the inclusion criteria. The full texts of potentially eligible articles were retrieved and assessed independently against inclusion criteria by two researchers (HT and LP or HM). Data extraction and quality assessment were conducted by one researcher (HT) and checked by a second researcher (LP).
We used the Cochrane risk-of-bias tool for randomised trials (Risk of Bias toolkit, version 2) to assess the quality of RCTs or quasi-RCTs that were included in either review update.20 Critical judgements were made of the following domains: randomisation process, deviation from the intended intervention, missing outcome data, measurement of the outcome and selection of the reported results. Each included study was judged to be at ‘low risk of bias’, ‘high risk of bias’ or to give rise to ‘some concern of bias’ in each of these domains. An overall study risk of bias was established using the following criteria:
low risk of bias – the study was judged to be at low risk of bias for all domains
some concern of risk of bias – the study was judged to give rise to some concerns in at least one domain, but not judged to be at high risk of bias on any domain
high risk of bias – the study was judged to be at high risk of bias in at least one domain, or the study was judged to give rise to some concerns for multiple domains in a way that substantially lowered confidence.
As in Marshall et al.,5 we used the quality assessment tool developed by Horner et al.18 to assess studies using single-case experimental design. This tool grades the presence of absence of 21 quality indicators within single-case experimental designs relating to (1) the description of participants and settings (three criteria), (2) dependent variable(s) (five criteria), (3) independent variable(s) (three criteria), (4) the establishment of a robust baseline measure and replicable description of the intervention (two criteria), (5) experimental control/internal validity (three criteria), (6) external validity (one criterion) and (7) social validity (four criteria). Each criterion is rated as met or unmet, giving a possible total of 21 per study.
Results
The numbers of references included and excluded at each stage of the study selection process for both review updates are shown in the PRISMA flow diagram ().
Study selection PRISMA flow diagram. M, Marshall et al.: N, NICE.
Following de-duplication, 316 references remained for assessment against the two sets of review inclusion criteria. Of these, 286 references were excluded based on the information in the title and abstract. The remaining 30 references were then assessed based on the full-text publication. Of these, a further 24 were excluded from the update review. A total of six references, reporting six studies, met the inclusion criteria adapted from Marshall et al.5 and NICE6 and were included: four references and two references, respectively (see ).
Update of the Marshall et al.5 review focusing on children with autism spectrum disorder
Four studies were included in our update of the Marshall et al.5 review (Box 3): two RCTs (one with a waiting list control21 and one with an active control22), one single-case experimental design23 (also referred to as a ‘n of 1’ study) replicated across participants and one pre–post pilot trial.24 Nine single-case experimental design studies were excluded because they included fewer than five participants. These nine studies were included in the mapping review (see Chapter 4). Full descriptive details of the included studies are provided in the data extraction tables (see Appendix 3); however, a brief summary is provided here.
Primary studies included in our update of Marshall et al.
The total population across all four studies was 107 children aged 3–8 years with a diagnosis of ASD. Three of the four studies included training interventions delivered to parents to address their children’s food aversions, restricted diets or mealtime behaviour.21,22,24 Peterson et al.23 delivered caregiver training after the study, whereas Johnson et al.,24 Marshall et al.22 and Sharp et al.21 incorporated training of the caregiver/parent by the therapist in a clinic setting, either from the beginning of the study or incrementally as the child/parents progressed. All of the studies included a behavioural intervention, using prompts (verbal and visual) and reinforcement of the child’s behaviour. Both Marshall et al.22 and Peterson et al.23 included a comparison intervention, with Graded exposure to food through modelling and play. In both studies the comparison intervention was delivered by the therapist; however, Marshall et al.22 incorporated training of the caregiver/parent, whose involvement incrementally increased as the sessions progressed. The number and duration of sessions varied across the studies for both interventions and comparators, with one22 offering the choice of sessions delivered weekly for 10 weeks, or sessions delivered intensively within 1 week. The duration of follow-up ranged from 8 to 16 weeks. The most common target outcome was mealtime behaviours, both positive and disruptive. Other outcomes measured across the four studies comprised Dietary intake, Dietary variety, Food acceptance, Mouth clean, Grams consumed, Weight, Height, Body mass index (BMI), Behaviour outside mealtimes, Parent stress, Caregiver satisfaction and Feasibility.
The overall quality of the studies that were included in the update was rated as low. The two studies that used a single-case experimental design both scored highly on baseline and external validity criteria and poorly on social validity; both failed to meet two of the four elements of this criteria [(1) magnitude of change of dependent variable is socially important and (2) independent variable is practical and cost-effective]. In addition, Johnson et al.24 scored poorly on the criteria description of participants/setting. The two RCTs21,22 were both considered to be at high risk of bias, with the main concerns being the measurement of outcomes and deviation from the intended interventions. Summaries of the quality assessments are presented in and (see also ).
Quality assessment summary from our update of the 2015 Marsh et al. review
Quality assessment of RCTs using the Cochrane risk-of-bias tool, version 2. –, high risk of bias; +, low risk of bias; ?, unclear risk of bias.
Quality assessment summary from our update of the 2017 NICE review
Summary of the overall evidence (from Marshall et al.5 and our update)
Twenty-three studies were included in the report by Marshall et al.;5 adding the four studies included in the update, the total number of studies was 27, two of which were RCTs. In Marshall et al.,5 only 5 out of the 23 (22%) studies reported an increase in the number of foods eaten, and only two studies used a formal outcome measure to capture this information. The four studies that were included in our update found a similar pattern, with only one study reporting outcomes relating to improvements in Total number of foods consumed and Total number of unprocessed fruits and vegetables.
In Marshall et al.,5 all studies were appraised using the quality rating tool for single-case experimental designs:18 scores ranged from 7 to 18 (out of a target total of 21). The two studies21,22 identified in the update that were assessed using this tool were generally consistent with Marshall et al.,5 but scored in a higher range (range 15–18). In their original review, Marshall et al.5 had called for prospective randomised trials to further demonstrate experimental effect, and one RCT22 and one pilot RCT21 were included in the update. Sharp et al.21 purported to represent the first RCT of a feeding intervention in ASD; however, this was a pilot RCT and, therefore, inadequately powered. Furthermore, this study21 was not sufficiently robust to allow strong conclusions to be made on the effects of EDSD interventions to improve the mealtime behaviour or foods eaten by children with ASD, or to inform decision-making. Overall, although the number of children with ASD who have taken part in research on interventions for EDSD has increased (from 37 children in 2014 to 144 children in 2018), the evidence base remains very limited. Although one RCT and one pilot RCT have now been included in the review, these studies contain methodological limitations. There continues to be a lack of rigorous studies and no high-quality prospective randomised trials to guide practice.
Update of the National Institute for Health and Care Excellence cerebral palsy in under-25s review6
Two new studies were included in our update of the NICE review6 to inform the management of EDSD in children with cerebral palsy.25,26 One was a RCT25 and the other was a small pilot study26 in preparation for a RCT. Full descriptive details of the included studies are provided in the data extraction tables (see Appendix 3); however, a brief summary is provided in Box 4.
Primary studies included in our update of the 2017 NICE review
The population of the two studies (combined, n = 100) was children who were aged 1.1–8 years with cerebral palsy. Serel Arslan et al.25 evaluated a parent training intervention that was delivered by speech and language therapists, whereas Song et al.26 piloted a therapist-delivered intervention. Both studies included interventions with multiple components that were carried out simultaneously and both included Positioning and Oral/sensory desensitisation.
Serel Arslan et al.25 (n = 80 children) evaluated an intervention that comprised Functional chewing training, which was made up of five steps: positioning the child, positioning food, sensory stimulation, chewing exercises and adjustments to food consistency. This was carried out alongside Oral motor exercises: five sets per day, five days per week, for 12 weeks. In addition, parents were given a brochure on exercises.
Song et al.26 compared Oral sensorimotor treatment (10 minutes) and neuromuscular electrical stimulation (20 minutes) twice weekly for 8 weeks with Oral sensorimotor treatment plus sham neuromuscular electrical stimulation in a pilot study. Oral sensorimotor treatment included various sensory stimuli that were applied to the cheeks, chin, lips, tongue and oral palate using human fingers, a vibrator and an ice stick.
Both studies measured Feeding behaviour as a primary outcome, but used different tools to evaluate progress. Serel Arslan et al.25 also measured Chewing function, whereas Song et al.26 measured Severity of dysphagia. The study by Song et al.26 was a pilot study and as a consequence it was judged to be at high risk of bias, whereas the full study by Serel Arslan et al.25 was considered to be at lower risk of bias. A summary of the quality assessments are presented in .
Summary of totality of the combined evidence (National Institute for Health and Care Excellence’s review6 plus our update)
Eight studies were included in the original NICE6 review of primary management of EDSD, and two more were included in our update (total number of included studies, n = 10; total number of participants, n = 335). The review included six RCTs and four cohort studies of various interventions and the studies were conducted in a number of countries: four in the USA, two in Turkey, and one each in the Republic of Korea, Canada, Bangladesh and the Islamic Republic of Iran, potentially limiting generalisability to the NHS setting. Based on its original review, NICE made a number of guideline recommendations (see Box 2). The combined evidence, including our updated review, does not support any changes to those guidelines.
Update of the review of oropharyngeal dysphagia by Morgan et al.4
Searches and data extraction for an update of the review by Morgan et al.4 were conducted by the original review team in 2018. An updated review is due to be published in 2020. Inclusion criteria for this review are:
Population – children and young people aged < 18 years with oropharyngeal dysphagia (i.e. difficulties in chewing or preparing food, moving food posteriorly with the tongue and swallowing food) diagnosed by a medical officer. Studies involving children with oesophageal dysphagia, including lower oesophageal sphincter dysfunction and gastro-oesophageal reflux, were excluded.
Intervention – any intervention that aimed to improve body functions underpinning eating/drinking, eating/drinking or participation in mealtimes.
Outcomes – Physiological Function, Aspiration and chest health, Diet consumed, Growth, Participation at mealtimes and Parental stress.
Study design – RCTs and quasi-RCTs.
Language – studies were published in English in peer-reviewed journals.
The review by Morgan et al.4 included three studies (two involving children with cerebral palsy, both of which were included in the NICE review,6 and one additional study that focused on children with myotonic dystrophy). In their recent searches, Morgan et al.4 identified one additional study: Sığan et al.27 This study was a RCT and included 81 children with cerebral palsy who were aged 12–42 months. At the time of this report, Morgan et al.4 had not extracted data from this paper or reviewed its quality. However, it was included in the original review by NICE6 and, therefore, has already been systematically appraised; it was rated as being low quality in the NICE review.6
Summary of systematic review findings
The original published systematic reviews of interventions and our updates have demonstrated that the evidence to address the questions around effective management of EDSD in children with cerebral palsy or ASD is of low quality and is accumulating slowly. A number of pilot RCTs were identified and included, some of which drew inferences around effectiveness (albeit with a lack of power); therefore, despite their primary aim being around feasibility, we have included them for completeness. Drawing the evidence together in one report has allowed us to identify the overlap in studies reviewed and some of the interventions that are being developed to improve children’s outcomes. Most RCTs involving children with cerebral palsy have focused on Sensorimotor treatments, with Behavioural techniques and parent training in Positioning, Modification of equipment, Food, and Environment and mealtime management. Interventions for children with ASD have used Behavioural techniques to improve mealtime behaviour and food aversion. A wide variety of outcomes have been targeted, using many different outcome measures. Meta-analysis has not yet been possible in any of the three reviews; no attempt was made to update the original, novel, meta-analysis by Marshall et al.5 There remains a lack of high-quality studies and there are no high-quality prospective clinical trials that demonstrate the effectiveness on EDSD interventions for young children with neurodisability. We conclude that there is inadequate research evidence to demonstrate whether interventions to improve EDSD in children with cerebral palsy or ASD are effective.
Strengths and limitations of systematic reviews
To make best use of these published systematic reviews of interventions we updated each using the methodology of the original review, with the exception of the age criterion. To align with our project aims we extended the age criterion to 8 years for all of the review updates. By following each of the review methods different risk-of-bias and assessment tools were used across the updates to appraise the specific quality markers of RCTs and single-case experimental designs. Although the tools for assessment differ, resulting in different presentations, the criteria against which risk of bias and quality are being assessed are generally the same. We have assumed that the original searches and processes were robust enough to identify all relevant studies, as each of the reviews followed established systematic review processes. Each review includes and represented the best available evidence (with the inclusion of experimental designs) around effective management of EDSD in children with cerebral palsy or ASD. Our updates have ensured that the findings and recommendations of these reviews remain up to date.
Patient and public involvement in systematic reviews
The PAG considered the summaries of the findings from the updated systematic reviews of interventions alongside findings from the mapping review and national survey. The PAG advised on creating a pictorial summary of the identified interventions and outcomes to aid discussion in the second focus groups and the stakeholder consultation workshops (see and ).
How did the systematic review findings inform the next step?
The interventions and outcomes identified in the updates of the published systematic reviews of interventions were considered by parents and professionals in the first focus groups with regard to their use in the UK (see Chapter 5). They also directly informed the design of the national survey of current practice (see Chapter 7) and the outcome measures used in studies included in the updates of the three published systematic reviews of interventions were listed for inclusion in the systematic review of measurement properties of tools (see Chapter 6).
