Feeding difficulties in children with food allergies: An EAACI Task Force Report

The term “feeding difficulties” refers to a spectrum of phenotypes characterized by suboptimal intake of food and/or lack of age-appropriate eating habits. While it is evident that feeding difficulties are prevalent within healthy children, no consensus has been reached for those with food allergies. The aim of this study was to systematically review all the available literature reporting the prevalence of feeding difficulties within food allergic children. We searched eight international electronic databases for all published studies until June 2022. International experts in the field were also contacted for unpublished and ongoing studies. All publications were screened against pre-defined eligibility criteria and critically appraised by established instruments. The substantial heterogeneity of included studies precluded meta-analyses, so narrative synthesis of quantitative data was performed. A total of 2059 abstracts were assessed, out of which 21 underwent full-text screening and 10 studies met the study criteria. In these, 12 different terms to define feeding difficulties and 11 diagnostic tools were used. Five papers included data of feeding difficulty prevalence in children with food allergies, ranging from 13.6% to 40%. Higher prevalence was associated with multiple food allergies. The current literature suggests that feeding difficulties are prevalent within food allergic children, particularly those with multiple food allergies. However, the heterogeneity of terminologies and diagnostic tools makes drawing conclusions challenging. Consensus guidelines for the diagnosis and management of feeding difficulties within food allergic children and further research

against pre-defined eligibility criteria and critically appraised by established instruments.The substantial heterogeneity of included studies precluded meta-analyses, so narrative synthesis of quantitative data was performed.A total of 2059 abstracts were assessed, out of which 21 underwent full-text screening and 10 studies met the study criteria.In these, 12 different terms to define feeding difficulties and 11 diagnostic tools were used.Five papers included data of feeding difficulty prevalence in children with food allergies, ranging from 13.6% to 40%.Higher prevalence was associated with multiple food allergies.The current literature suggests that feeding difficulties are prevalent within food allergic children, particularly those with multiple food allergies.However, the heterogeneity of terminologies and diagnostic tools makes drawing conclusions challenging.Consensus guidelines for the diagnosis and management of feeding difficulties within food allergic children and further research

| INTRODUC TI ON
Feeding difficulties is an umbrella term encompassing a spectrum of phenotypes, characterized by suboptimal intake of food and/or lack of age-appropriate eating habits. 1 Classifications of these phenotypes have been suggested in the literature, based upon three principal concerns of parents: eating too little, selective intake, and fear of feeding. 2,34][5][6][7] Feeding disorders can have long-lasting adverse impacts beyond childhood on various on the development and perpetuation of feeding difficulties are needed to appropriately manage such patients.

K E Y W O R D S
eating difficulties, feeding difficulties, food allergy, prevalence, systematic review

G R A P H I C A L A B S T R A C T
Graphical abstract summarising the key findings of this systematic review.

Key message
Feeding difficulties are prevalent within food-allergic children, particularly those with multiple food allergies.There is no consensus on how to assess feeding difficulties in food-allergic children and great heterogeneity of definitions and diagnostic criteria was found to assess feeding difficulties.Future work should focus on developing such tools to harmonize clinical outcomes and advance the field through prospective research.
4][15] In children with food allergies, the limited exposure to different flavors and textures due to the elimination diet 16 limits the development of the appropriate oral-motor and sensory functions needed for feeding. 7,17,18termining the true prevalence of feeding difficulties is challenging due to the heterogeneous nature of the pediatric population and the lack of consensus regarding terminologies, definitions, and diagnostic criteria.A significant proportion of healthy young children experience periods of food refusal as they become more autonomous and food neophobia is part of the normal development of all omnivores. 17,19Available research indicates that feeding difficulties are seen in 25%-45% of the general pediatric population, in 80% of children with developmental disabilities, and in 40%-70% of children with chronic medical conditions. 7The occurrence of feeding difficulties within food-allergic children is also becoming increasingly recognized, but reported ranges, using different feeding difficulty terminologies, vary significantly. 16,20,21ile guidelines exist for the management of feeding difficulties and disorders in the general pediatric population, 3,7 there are no consensus guidelines specific for those with food allergy, which may lead to misdiagnosis and mismanagement.[24] In this work, we propose the first systematic review of all the available literature on feeding difficulties within food-allergic children and adolescents (up to 18 years of age) in order to determine whether the prevalence is higher than that in the general population.
TA B L E 1 Glossary of terms for the feeding difficulties included in this systematic review.

Feeding difficulty Definition
Adaptive feeding Caregiver use of different strategies to improve the child's nutritional status; these can be functional or maladaptive Avoidant/restrictive food intake disorder (ARFID) A recognized feeding disorder since 2013.

| Study eligibility
The PICOS 26 framework was used to design the study eligibility criteria as follows: healthcare professional (HCP) were included.Publications involving children with eating disorders (e.g., anorexia nervosa) were excluded, as were papers on cohorts of children with organic disorders that have been linked to high incidences of feeding difficulties, for example, autism spectrum disorder (ASD).Publications focusing on eosinophilic oesophagitis (EoE) were syphoned to be reviewed separately in order to distinguish papers that described feeding difficulties as opposed to dysphagia and food impaction, which are direct presentations of the disease.The assessment of feeding difficulties in EoE is currently underway as a separate publication in a follow-up to this publication.

| Interventions/conditions
All types of feeding difficulties were included in this review.

| Outcome
The primary outcome of this systematic review was to establish the prevalence of feeding difficulties (Table 1) within the population of children with food allergies; therefore, studies that did not include quantitative prevalence data were excluded.The prevalence and any definitions and diagnostic criteria of the feeding difficulties were noted.Any impacts of feeding difficulties on growth, HRQoL, mental health, and school/work absenteeism in children and/or their parents, if investigated, were also reported.

| Study design
All types of studies: randomized-controlled, non-randomized, crosssectional, case-controlled, cohort, and case series (defined as five or more case reports) were included.Animal studies, review papers, case reports, studies in abstract-form only, and qualitative papers were excluded.

| Screening of studies
Two independent reviewers (SH, UN) first screened the abstracts of retrieved articles, followed by the full text of potentially relevant papers.Any discrepancies were resolved by consensus and a third reviewer (RM) arbitrated any disagreements at each stage.Studies referencing food intolerances were screened beyond the abstract to determine whether they were describing non-IgE-mediated food allergy and were to then be included.

| Data extraction and reporting
Two reviewers independently extracted relevant information and study data onto a customized data extraction sheet (Appendix S2).
Descriptive tables were used to summarize the literature and characteristics of studies contributing to the overall evidence.

| Quality assessment and risk of bias
Two reviewers (SH, UN) independently assessed the methodological quality of eligible studies and the potential for risk of bias using the Effective Public Health Practice Project (EPHPP). 27Any discrepancies, if encountered, were arbitrated by a third reviewer (RM).
Overall grading for each observational study as well as componentspecific measures were assessed, including suitability of the study design for the research question; risk of selection bias; and outcome assessment.

| Data syntheses
All data were qualitatively analyzed.

| Search results
A preliminary search of all eight databases yielded 2949 potentially eligible papers.After the removal of duplicates and screening the abstracts of the remaining publications, 21 papers underwent full-text screening (Figure 1).8][39][40][41][42][43][44] We found no interventional studies in this systematic review.The 10 observational studies comprised of two retrospective medical chart reviews, four cohort and four crosssectional case-control studies.
The results of this systematic review have been divided into two sections: five papers including absolute values of the prevalence of feeding difficulties and five papers including quantitative data only on the mean/median feeding difficulty scores of a cohort of food allergic children.

| Characteristics of included papers
Across the 10 analyzed publications, 11 different diagnostic tools and 12 different terminologies of feeding difficulties were used (Tables 2   and 3).Some of the included studies assessed multiple feeding difficulty phenotypes and therefore used different terminologies.Seven of these publications used only one diagnostic tool, two of which both used the CEBQ, therefore accounting for six of the different diagnostic tools reported in this systematic review.Despite only one diagnostic tool being used, only four of these papers used one feeding difficulty term, whereas Maslin et al. 41 used two, and Polloni et al. 39 TA B L E 2 The characteristics of the five studies that contain prevalence data on feeding difficulties within food-allergic children.used three different terminologies, respectively.Maslin et al. 9 used three different feeding difficulty terms, while Rigal et al. 44 used only one, despite both using two different diagnostic tools.Rodrigues et al. 38 used three different diagnostic tools and four different feeding difficulty terms.Tables 2 and 3 show the characteristics of the studies containing prevalence and quantitative data, respectively.

| Quality appraisal of included studies
The global quality rating of the studies considered five studies to be strong, one moderate and four weak (Table S2).Of these four weak studies: two were retrospective chart reviews, one of which disregarded confounders, and one had no validated data collection method; two were case-control studies with a high number of unexplained patient dropouts and either no mention of confounders or a high selection bias.

| Prevalence data
Five of the included studies 9,16,[37][38][39] reported the prevalence of feeding difficulties in children with food allergies, ranging from 13.6% to 40% (Table 4).Only one of these five papers focused on a cohort of children with IgE-mediated food allergy, 39 recording a 19% prevalence of feeding difficulties.Out of the remaining four papers: two focused on cohorts of children with non-IgE-mediated food allergies and reported a prevalence of feeding difficulties ranging between 21.7% and 30%; two focused on children with Cow's Milk Allergy (CMA), the underlying etiology of which was not included, recording the prevalence of feeding difficulties to range between 13.6% and 40%.
Rodrigues et al. 38 reported the prevalence of three different feeding difficulty phenotypes: 35.4% prevalence of picky eating in the CMA group, which comprised of children with both IgE and non-IgE-mediated CMA, compared to 23.3% of their control group; 32.1% prevalence of feeding problems, which was non-significantly different than their control group (28.4%), and 23.9% prevalence of avoidant eating behavior.Conversely, using the same questionnaire to investigate the latter two forementioned phenotypes, Maslin et al. 9 reported 13.6% prevalence of feeding difficulties, compared to 1.6% in their control group, and 40% prevalence of avoidant eating behavior in their CMA group, which also included children with both IgE and non-IgE mediated CMA; these were both the lowest and highest reported prevalence of feeding difficulties found in this systematic review.
Two of the papers were retrospective chart reviews of non-IgE mediated food allergies, with patient records sourced directly from tertiary clinics. 16,37The reported frequency of the respective feeding difficulties ranged from 21.7% to 40.2%.
Meyer et al. 16 (UK) retrospectively reviewed 437 cases of non-IgE-mediated allergies.40.2% of parents recalled avoidant eating behaviors; 75% of these cases were supported by medical records, concluding a 30% prevalence.Su et al. 37 (USA) reported food aversion in 21.7% of Food Protein-Induced Enterocolitis Syndrome (FPIES) cases, with no restriction on diagnostic criteria.TA B L E 3 The characteristics of the five studies that contain quantitative data on the feeding difficulty scores of a cohort of food allergic children.

| Quantitative data
Five of the included papers [40][41][42][43][44] report quantitative data relating to feeding difficulties in food allergic children (Table 5).Again, only one of these five studies focused on a cohort of children with IgE-mediated food allergies, 42 and reported significantly increased scores assessing behavioral feeding difficulties (BPFAS), in their food-allergic group compared to healthy controls.
Two studies investigated feeding difficulties (CEBQ) in CMPA children. 41,43Maslin et al. 41 (UK) reported a 7.8% increase in prevalence of feeding difficulties when comparing the CMPA and control group, whereas Ercan and Tel 43 (Turkey) reported mixed results.
Mixed results for food neophobia were also found: Rigal et al. 44 reported significantly increased scores in food-allergic children when compared to their non-allergic siblings, whilst Maslin et al. 41 reported non-significant differences in scores between the CMA and non-allergic control group.

| Associations
Four papers 16,37,39,41 reported an increased association with the presence/severity of feeding difficulties and the number of foods eliminated from the diet.Sub-group analysis by Su et al. 37 showed the prevalence of food aversion to be significantly different in those eliminating one/two foods, due to acute FPIES, compared to three or more: 16.9% and 43.2% respectively.One of the included papers found a younger patient age to be associated to stronger avoidant eating behaviors. 38The number of food allergy symptoms, including vomiting, diarrhea and constipation 16,38,41 and extra-intestinal manifestations such as headaches, night sweats, lethargy and joint pain 16 and colic and dry cough at night 9 were also associated to increased prevalence/severity of feeding difficulties.A family history of food allergy was also associated with an increased food aversion score. 37,38Parental trait and state anxiety scores were also found to negatively correlate to the change in nutritional habits after an oral food challenge (OFC). 39 associations between country of study, race, gender, or study design and the prevalence of feeding difficulties were made apparent.

| Impacts
Two of the included studies which focused on children with CMA showed the long-term effect of cow's milk elimination diets within the first 2 years of life on persisting avoidant feeding behavior 7-10 years later, 41 and altered nutritional habits at age 2-6, 43 respectively.

TA B L E 4
]45 First author and year of publication Size of study The 75th percentile score of the control group was used as a cutoff score for picky eating in the food allergic group.
One study reported unaffected growth, 9 whereas four studies reported either a higher number of underweight children and reduced growth, 41 weight loss/poorer weight gain 16,37 or failure to thrive 45 in the presence of feeding difficulties.Anxiety in the patient and anxiety or stress in the parents were also reportedly higher in those affected by feeding difficulties. 13,39,45

| DISCUSS ION
To the best of our knowledge, this is the first systematic review on the prevalence, terminology (definitions), and outcomes of feeding difficulties in food-allergic children.This review demonstrates the sparsity of publications and lack of consensus on terminology and diagnostic criteria specifically in relation to food allergy.
The primary objective was to establish the prevalence of feeding difficulties in food-allergic children.Across six studies, using a variety of criteria, this prevalence ranged from 13.6% to 40%.This increased to 40.2% when including parental recall reports of feeding difficulties; however, studies have shown the lack of accuracy with parental recall of dietary habits in children. 44thin the literature, with the use of different diagnostic questionnaires and criteria, the prevalence of feeding difficulties in healthy children has been reported in the range of 20%, 20 25%-30% 3,46 and even up to 45%. 7,47Picky eating in healthy children has been reported to range between 14% and 50%. 19,48As a result of this significant range of prevalence (14%-50%), it is challenging to ascertain which figure should be referenced when comparing healthy and food-allergic children.
Maslin et al. 9 and Rodrigues et al. 38 used the same diagnostic tool in children with CMPA but reported a prevalence of 13.6% and 32.1% of feeding difficulties/problems, respectively.This may be explained by the difference in the proportion of children eliminating two or more foods: 28.8% in the study by Maslin et al. 9 compared to 63% in the study by Rodrigues et al. 45 Also, while the age of food allergy onset was similar, the age at time of study was significantly different: 13 months and 3.3 years.Two published studies included in this review indicated a peak prevalence around age 3, 41,43 which has previously been hypothesized in the literature. 38,49,50It is worth noting, however, that none of the included papers reported the age of feeding difficulty onset or diagnosis.
Out of the two retrospective chart reviews, Meyer et al. 16 reported the highest prevalence of feeding difficulties (30% as assessed by HCPs) in a population of non-IgE-mediated allergic children.The patients in this study were recruited from a tertiary referral center that specializes in gastroenterology, with 78.7% of included children eliminating three or more foods, as opposed to only 18.2% in the retrospective analysis by Su et al. 37 (21.7% prevalence).The predominance of acute FPIES (88.7%) in the review by Su et al., 37 compared with the more diverse cohort of non-IgE-mediated allergies in the review by Meyer et al., 16 may also contribute to this difference in prevalence, with chronic phenotypes more strongly associated with learned feeding aversions.
The study by Meyer et al. 16 may therefore be representative of a specific, more severe, allergic phenotype.In addition, while the Wright et al. 20 criteria had been used in a healthy UK population before, behaviors such as closing mouth or turning head when food is offered are commonly seen in normal developing toddlers,

Means of measuring feeding difficulty
Feeding  This number refers to published data from typically developing children.
thereby potentially contributing to an over-reporting of feeding difficulties by Meyer et al. 16 Repeated, extended episodes of discomfort and pain from feeding are often experienced by patients with non-IgE-mediated food allergy for an extended period of time as a result of delayed trigger identification or lack of symptom resolution despite food avoidance. 51Non-IgE-mediated food allergies are therefore thought to have a stronger association with feeding difficulties than IgE-mediated allergies.Only one identified paper reported the prevalence of feeding difficulties in IgE-mediated food-allergic children (19%).Only 28.4% of the assessed cohort had more than two food allergies.In addition, many different feeding difficulties, such as food refusal, selective eating, food neophobia and lack of appetite, as well as parental misperception, could have contributed to the "lack of interest in new foods" and "monotony of diet" that was reported by their own investigative tool.It is therefore difficult to compare the results of this study to others.One of the other included papers reported the behavioral feeding difficulty scores 52 of IgE-mediated food allergic children with feeding difficulties, which, although was found to be statistically higher than that of typically developing children, did not compare to children with non-IgE-mediated allergy. 42This systematic review is therefore unable to conclude any specific comparative data on non-IgEand IgE-mediated allergies.
Food allergy management relies on strict adherence to elimination diets and emergency treatment in the event of allergen exposure when reactions occur.Four of the studies reported an association between increased number of eliminated foods and the presence and severity of feeding difficulties. 16,37,39,41This, coupled with the long-term persistence of feeding difficulties shown in children up to 10 years after food allergy remission, 41,43 highlights the potential effects of the elimination diets themselves or the resultant anxiety or fear from a perception of greater risk.
Further examples of this have been described in case reports in the literature. 21e of the secondary objectives of this systematic review was to identify the different definitions used for feeding difficulties.
Out of the 10 included studies, 11 different diagnostic tools and 12 different terminologies for feeding difficulties were used (Table 1).
Different terminology and diagnostic criteria have been used in the literature dependent on the type of food allergy, and therefore account for the expected direct presentations of the disease, feeding difficulty phenotype and the age groups being assessed, due to different expected developmental milestones with regard to oralmotor skills and self-feeding ability.Multiple tools exist for assessing feeding difficulties in pediatrics but consensus in the literature for the best validated tools for food-allergic children, stratified by population subsets, does not exist.The studies included in this systematic review therefore use criteria that either the authors themselves had used before, either in clinical practice or research, or had been used in similar populations in the literature.It is therefore not possible, based on the publications included in this systematic review, to suggest a most appropriate specific terminology and specific tool for the assessment of feeding difficulties in food-allergic children.
However, Godday et al. 7 published a consensus definition and diagnostic criteria for pediatric feeding disorder in 2019, which may be useful in future studies assessing feeding difficulties in food allergic children.
Furthermore, this systematic review also aimed to assess the impact of feeding difficulties.the heterogeneity of results due to their vastly different clinical presentations.Also, while all studies contained children with food allergies diagnosed by a HCP, oral food challenges, the gold standard for diagnosis, were performed on all patients in only two studies. 43,44ile Ercan et al. 43 and Maslin et al. 41 have suggested that there is a persistence of the feeding difficulty phenotype many years after the resolution of food allergy, none of the included studies contained long-term follow-up data on the therefore only be representative of the more challenging end of the food allergy spectrum.This suggests that feeding difficulties might be more prevalent in specific subsets of food-allergic children.
The increased parental anxiety 3,8,13,39 and attention on feeding and mealtime behavior 11 in parents of food allergic children is likely to lead to an inherent bias in the reporting of feeding difficulties.
The amplified state of vigilance, to avoid accidental food allergen exposure, has been shown to increase within food allergic children as they age. 10,59Misconceptions among parents and HCPs on what constitutes "normal" feeding development, and which feeding difficulties require further evaluation, monitoring, and management, may have influenced the data.What may be regarded as dysfunctional feeding, may actually be necessary adaptation to living with food allergy.

| Strengths of the study
The comprehensive review of eight international electronic databases with high methodological rigor increases the strength of the conclusions of this systematic review.
Future studies that could address the knowledge gaps identified in this systematic review are summarized in Appendix S3.

| CON CLUS ION
This systematic review supports the idea of feeding difficulties being common in food-allergic children, particularly those with multiple food allergies.Great heterogeneity in definitions and diagnostic criteria was identified.Given the increasing prevalence of food allergies, this highlights the need for consensus-based definitions and diagnostic tools for feeding difficulties in food allergy to ensure early recognition and optimal management by multidisciplinary teams.This Task Force aims to conduct a Delphi Consensus exercise to reach agreement on which tools and terminology should be used to assess feeding difficulties in children with food allergies.
Prospective, long-term follow-up data in this area are needed to understand long-term patterns as well as the potential underlying pathologic mechanisms and risk factors linking food allergy to the development of feeding difficulties.

2. 2 | 5 of 13 HILL
.1 | Population Studies of children and adolescents up to 18 years of age with IgE, non-IgE, or mixed IgE and non-IgE-mediated food allergy, diagnosed by a F I G U R E 1 PRISMA flow diagram of screening and selection of studies for qualitative analysis. 25PRISMA methodology was used to guide the reporting of this systematic review.PRISMA, Preferred Reporting Items for Systematic Reviews and Meta-analyses.et al. c

| 11 of 13 HILL
development and evolution of feeding difficulties within this population.The high number of cross-sectional studies in this review therefore does not allow us to exclude transient feeding difficulties, such as selective eating, food refusal, or food neophobia phenotypes, which are often seen in healthy children.2Information on supportive interventions such as dietician access was also not included in any of the publications.Therefore, the potential effects such support may have had on the development and/or persistence of feeding difficulties also cannot be assessed.Nine of the included studies relied on direct tertiary clinic recruitment of patients, which may indicate a more complex, severe, or persistent profile warranting such referral.These studies may et al. 9,16,[36][37][38]45 Abbreviations: CMPA, Cow's Milk Protein Allergy; FPIES, food protein-induced enterocolitis syndrome.aThreeor more food allergies.| 7f 13HILL et al. 13,[39][40][41][42][43] Abbreviations: BPFAS, Behavioural Paediatrics Feeding Assessment Scale; CEBQ, Child Eating Behaviour Questionnaire; CMPA, Cow's Milk Protein Allergy; FPI, Food Preference Inventory; FSQ, Food Situation Questionnaire.
8% increase in prevalence of avoidant eating behavior in the food allergic group compared to healthy control group; no absolute prevalence values available.