Risk factors for avoidant/restrictive food intake disorder in children: A systematic review

Abstract

Objectives

Avoidant/restrictive food intake disorder (ARFID) is a relatively new diagnosis in the Diagnostic and Statistical Manual of Mental Disorders (DSM)‐5, since 2013. The restrictive and/or selective eating—driven by a lack of interest, sensory sensitivity, and/or concern over aversive consequences—is associated with significant medical and/or psychosocial problems. However, little is known about risk factors of this feeding/eating disorder. Our objective was to investigate which factors predispose to ARFID in children.

Methods

Databases CINAHL, Embase, PsycINFO, PubMed, and Web of Science were searched to identify peer‐reviewed articles published from inception to August 27, 2024. Articles that included children with ARFID, a control group, and a potential risk factor were selected.

Results

A total of six studies were included in this review. In these studies, data regarding potential risk factors and their risk of bias were evaluated. All studies were of cross‐sectional design. Despite an extensive search strategy, only a small number of studies could be included for analysis. A total of 18 factors were reported, of which 10 factors were significantly associated with ARFID. These factors include physical factors, psychiatric comorbidities, and maternal psychopathological risk.

Conclusions

This systematic review shows which factors are associated with ARFID in children. To better clarify the direction of associations between potential risk factors and ARFID, longitudinal and interventional studies are needed. This knowledge may support early recognition and timely intervention in ARFID.

What is Known

• Since the establishment of its diagnosis in 2013 in the Diagnostic and Statistical Manual of Mental Disorders (DSM)‐5, several hypotheses were formed about the development of avoidant/restrictive food intake disorder (ARFID).

• Little is known about specific factors that may predispose a child to developing ARFID.

What is New

• There are several factors proven to be associated with ARFID in children, such as comorbid psychiatric disorder and maternal psychopathological risk.

• There is a lack of longitudinal research to better understand the temporal relationship between potential risk factors and ARFID, highlighting the need for future studies, including randomized controlled trials to determine causality.

1. INTRODUCTION

Feeding and eating problems are common in childhood. ¹ Avoidant/restrictive food intake disorder (ARFID) is characterized by significant medical and/or psychosocial sequelae that require clinical attention. ² Before its inclusion in the Diagnostic and Statistical Manual of Mental Disorders (DSM)‐5 (2013), many children with feeding or eating difficulties did not meet DSM‐IV criteria for anorexia or bulimia nervosa and were classified under eating disorder not otherwise specified (EDNOS) or feeding disorder of infancy or early childhood (in case of onset <6 years of age). This also included children with substantial feeding difficulties, such as chronic “picky eating” or “selective eating” leading to malnutrition, who presented with clinically significant impairment but did not meet the criteria for any established psychiatric eating disorder diagnosis. This gap in diagnostic recognition left many of these children without an appropiate medical or psychiatric classification. To enhance diagnostic clarity and clinical utility, the DSM‐5 introduced ARFID, replacing the previous categories of “Eating Disorders” and “feeding and eating disorders of infancy or early childhood”. ³ Prevalence estimates of ARFID vary between studies, ranging from 3.2% in primary school children and 5%–22.5% in children and adolescents with eating disorders. ³ , ⁴

Since ARFID is a relatively new diagnosis, research interest is growing. Potential risk factors for development of ARFID in children are not exactly known. ⁵ , ⁶ In previous studies, risk factors for ARFID have been identified in broad domains including somatic health, psychological factors (such as anxiety), sensory sensitivity and environmental influences such as feeding practices. ⁷ In addition to this, nonmodifiable risk factors such as sex and genetic vulnerability may also contribute to ARFID. ⁸

Because the disorder's representation is driven by one or more of three profiles, risk factors are assumed to relate these underlying drivers. ⁵ , ⁹ These profiles are not mutually exclusive, and individuals often show features of multiple types. The first profile, described as a lack of interest in eating or food, is often present from infancy. Potential risk factors include perinatal events such as prematurity, which is associated with illness and early tube feeding. ¹⁰ The second profile, involving sensory sensitivity for characteristics of food (e.g., smell, color, and texture), is often seen in children with autism spectrum disorder (ASD); the co‐occurrence of ASD may therefore increase the risk of developing ARFID. ¹¹ , ¹² The third profile, defined by fear of aversive consequences of eating, may arise after a traumatic experience such as choking, or from prior gastrointestinal discomfort or pain. ⁶

Identifying risk factors for ARFID is essential for early detection and informing timely intervention aimed at addressing nutritional deficiencies, undernutrition, and reducing associated health consequences. Accordingly, a systematic review was conducted to examine factors associated with ARFID in children. It should be noted that although the term ‘risk factor' is used, it refers to associated factors being potential risk factors, as causal relationships have not been confirmed through robust evidence. ¹³

2. METHODS

2.1. Ethics statement

This study did not require approval from an institutional ethics committee as it is a systematic review of previously published data and does not involve primary data collection or human participants.

2.2. Search strategy

Our search strategy was developed based on guidelines of the Cochrane Handbook and with the help of an information specialist, the full electronic search strategy for PubMed is included as Supplement S1. ¹⁴ In the initial PROSPERO registration, the search strategy included broad terms related to eating and feeding disorders and was limited to children up to 6 years old. However, the pilot screening yielded too many irrelevant records and few ARFID‐specific studies. To improve focus and feasibility, the strategy was refined to include only studies explicitly addressing ARFID, and the age range was expanded to all pediatric populations to ensure a more comprehensive synthesis. In November 2022, we conducted a literature search of the electronic databases CINAHL, Embase, PsycINFO, PubMed, and Web of Science using terms describing ARFID and risk factors. The search combined controlled vocabulary (e.g., MeSH, Emtree) and free‐text terms describing ARFID and risk factors. Example search terms included: “avoidant restrictive food intake disorder” OR ARFID OR “feeding disorder” AND “risk factor” OR predictor* OR etiology. The adjusted search strategy can be found in Supplement S2. The search was updated in August 2024 to identify additional studies, using the same strategy and databases. ¹⁵

The protocol was registered in the International Prospective Register of Systematic Reviews (PROSPERO CRD42022129603), and is reported using the Preferred Reporting Items for Systematic reviews and Meta‐Analyses (PRISMA) statement guidelines. ¹⁶

2.3. Eligibility criteria

For studies to be eligible for inclusion, they needed to fulfill the following criteria: children and adolescents (up to 18 years of age) being the participants, participants being diagnosed with ARFID according to the DSM‐5 definition  ² and diagnosis made by a health care professional, empirical, peer‐reviewed studies reporting original data (e.g., cohort, case‐control, cross‐sectional), the study should analyze at least one potential risk factor for ARFID such as sex or comorbid psychiatric conditions and include a comparison or control group. We restricted our inclusion to peer‐reviewed full‐text articles. Exclusion criteria were case reports, case series without a comparison group, narrative or systematic reviews, conference abstracts, and gray literature. Studies focusing on feeding problems outside ARFID's diagnostic criteria, or on cases where ARFID was secondary to a clearly identifiable medical condition or developmental disorder (e.g., severe neurological impairment, structural gastrointestinal disease), were excluded unless ARFID was examined as the primary diagnosis.

2.4. Screening and data extraction

Screening of all titles and abstracts was conducted independently in duplicate by two authors (R.N. and L.A.) using Endnote. Discrepancies were resolved through discussion, with arbitration by a third assessor if required. After resolving discrepancies, full‐texts of all articles were again screened independently and in duplicate for eligibility by the same two authors (R.N. and L.A.). The repeated search strategy (August 2024) was performed in the same manner (R.N. and M.K.). After final selection, the eligible articles were included and data were extracted using a standardized, piloted Excel template developed for this review. The following data from each article were extracted: risk factor(s) that could be associated with ARFID, sample size, mean age and range, sex, ethnicity, socio‐economic status, recruitment strategy, ARFID diagnosis (by whom and how) and study limitations (Table 1). Missing data from published articles were requested by contacting the study authors.

Table 1.

Study characteristics of included articles.

Author (year). Location	Investigated factor(s)	Study type	Sample size	Age ARFID (min‐max (mean ± SD)	Sex (% male)	Ethnicity	Recruitment strategy	ARFID diagnosis	Limitations
Canas et al. (2021). Spain	Medical and psychiatric comorbidity. Psychiatric and psychological antecedents. Attention and behavioral problems. Clinical fears: anxiety state/trait.	Cross‐sectional	33 ARFID 33 AN 33 control	7–17 years (10.85 ± 2.37)	60.6	ND	Sample drawn from eating disorder patients seeking treatment at hospital eating disorder unit, control group from general population. ASD was excluded	Semi‐structured interview using KSADS‐PL, complemented with a clinical interview, profession unknown	No details on reported medical and psychiatric antecedents small sample size
Cerniglia et al. (2020). Italy	Child emotional‐behavioral functioning. Maternal psychopathological risk. Mother‐child interaction	Cross‐sectional	73 ARFID 27 control	24–36 months (30 ± 3.07)	50.0	ND	Sample recruited through mental health clinics by psychologists	Independently by two clinicians (Cohen's k = 0.80)	No details on characteristics of the study groups small sample size self‐report measure maternal psychopathological risk
Cimino et al. (2021) Italy	DAT1 genotype and total methylation	Cross‐sectional	69 ARFID25 control	24–36 months (29 ± 3.14)	51.1	ND	Through public and private kindergartens, mental health clinics, and pediatric hospitals	Independently by two clinicians (Cohen's k = 0.80)	No details on characteristics of the study groups small sample size self‐report measure maternal psychopathological risk
Dovey et al. (2019). UK	Emotional problems. Non‐social sensory hypersensitivity	Cross‐sectional	29 ARFID 55 ASD 143 PE 255 control	Range months ND (101.4 ± 25.8)	53.5	90.2% European descent 6.8% mixed heritage 3% other 6.8% mixed heritage 3% other	Through an online platform	Reported diagnosis provided to the parent by a medical or psychological professional	No check on ARFID diagnosis all measurements by self‐report small sample size
Inoue et al. (2021). Japan	ASD	Cross‐sectional	32 ARFID 92 AN 6 unspecified feeding or eating disorder 496 control	5–15 years (11.8 ± 2.4)	21.9	ND	Multicenter cohort study 11 medical institutions	Psychiatrist or pediatrician and psychologist, direct observation and interview	Small sample size no gold standard ASD diagnosis across centers
Ye et al. (2023). China	Characteristics of gut microbes	Cross‐sectional	135 ARFID 33 control	ARFID 53–82 (66) months control 48–72 (60) months	48.1	ND	Through posters medical center	Structured clinical interview by pediatrician trained in diagnosing ARFID	Relatively small sample size, especially control

Note: Overview of included cross‐sectional studies on risk factors for ARFID in children, showing study design, investigated factors, sample details, recruitment, diagnostic method, and limitations.

Abbreviations: AN, anorexia nervosa; ARFID, avoidant/restrictive food intake disorder; ASD, autism spectrum disorder; DAT1, dopamine transporter 1; KSADS‐PL, kiddie‐schedule for affective disorders and schizophrenia for school‐age children‐present and lifetime; ND, not described; PE, picky eating; SD, standard deviation.

The risk of bias was assessed by two independent authors (R.N. and L.A.) using the Critical Appraisal Checklist for Analytical Cross Sectional Studies developed by JBI (Joanna Briggs Institute). ¹⁷ This is a 10‐item checklist with questions such as: “Were the criteria for inclusion in the sample clearly defined?,” which can be answered by “Yes,” “No,” “Unclear,” or “Not applicable.”

2.5. Statistical analysis

Due to the heterogeneity of studies, the results were considered to be not appropiate for pooling and performing a meta‐analysis. Therefore, a descriptive analysis was carried out and, where possible, Odds Ratios were calculated for the risk factors.

3. RESULTS

3.1. Study descriptive information and reporting

A total of 1765 records were identified, of which 876 appeared unique study reports after de‐duplication. ¹⁸ Of these, 82 reports were screened as full texts. A total of six studies were eligible for inclusion with a total number of 1.207 children. The study selection process is detailed in the PRISMA flow diagram, see Figure 1. ¹⁹ , ²⁰ , ²¹ , ²² , ²³ , ²⁴ Studies were conducted in five countries and study reports were published from 2019 to 2023 and all studies were of cross‐sectional design. For all study and subject characteristics and data extraction, see Table 1.

Figure 1.

PRISMA 2020 flow diagram of study selection process. The diagram summarizes the process of study identification, screening, and inclusion in the systematic review of risk factors for ARFID in children. The search (including the August 2024 update) yielded n = 1.765 records. After duplicate removal, n = 876 records were screened based on predefined inclusion and exclusion criteria, resulting in n = 6 studies included in the review. ARFID, avoidant/restrictive food intake disorder; PRISMA, Preferred Reporting Items for Systematic Reviews and Meta‐Analyses.

3.1.1. Age

All studies focused solely on pediatric populations. Two studies evaluated ARFID in young children of 24–36 months of age, ²⁰ , ²¹ the other four studies included children from 4 to 17 years. ¹⁹ , ²² , ²³ , ²⁴ Most studies reported age range, one study only reported the mean age and standard deviation (SD). ²²

3.1.2. Sex

Sex was reported in all studies. There was no report of the gender of the participants. Male sex percentages varied from 21.9% to 60.6%.

3.1.3. Ethnicity

Only one study reported on the participants' ethnicity. ²² In this study, 90.2% of the participants reported their ethnicity as a variant of European descent, 6.8% reported being of mixed heritage and 3% of African, Asian, South Asian, South‐East Asian, or Oceanic descent.

3.1.4. Study setting and recruitment strategy

Studies were conducted in five countries: Spain (n = 2), the United Kingdom (n = 1), Italy (n = 1), Japan (n = 1), and China (n = 1). According to the World Bank Group classification, four studies were conducted in high‐income countries (Spain, the United Kingdom, Italy, and Japan) and one in an upper‐middle‐income country (China). Recruitment of participants was reported to have taken place via mental health clinics (n = 2), ²⁰ , ²¹ medical institutions (n = 4), ¹⁹ , ²¹ , ²³ , ²⁴ and an online platform (n = 1). ²² Of these studies, one reported recruitment from both a mental health clinic and a medical institution. ²¹ Participants of the control groups were recruited from the general population without description (n = 1), ¹⁹ local junior high schools (n = 1), ²⁰ kindergartens (n = 1), ²¹ an online platform (n = 1), ²² a medical institution (n = 1)  ²⁴ and one study did not describe the recruitment method. ²³ The exact method of recruitment was only described in two studies. ²² , ²⁴

3.1.5. ARFID diagnosis

The ARFID diagnosis was made by clinicians (specialty not further described by the authors, n = 2), ²⁰ , ²¹ psychiatrist, pediatrician, and/or psychologist (n = 2). ²³ , ²⁴ One study (Dovey et al.) relied on parental report of a diagnosis made by a professional and Canas et al. did not report who made the diagnosis but it was assumed to be a specialized professional considering the location of recruitment was a hospital unit for eating disorders. ¹⁹ , ²²

3.1.6. Risk of bias

Risk of bias was assessed using the JBI checklist for cross‐sectional studies (Table 2). ¹⁷

Table 2.

Risk of bias assessment using the Joanna Briggs Institute critical appraisal checklist.

	Inclusion criteria clearly defined?	Participants and setting described in detail?	Valid and reliable exposure measurement?	Objective, standard criteria for measuring condition?	Confounding factors identified?	Strategies to deal with confounding factors?	Valid and reliable outcome measurement?	Appropriate statistical analysis?
Canas	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Cerniglia	Yes	No	Yes	Yes	No	Yes	Yes	Yes
Cimino	Yes	Yes	Yes	Yes	No	No	Yes	Yes
Dovey	Unclear	Yes	Yes	Yes	Yes	Yes	No	Yes
Inoue	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Ye	Yes	Yes	Yes	Yes	Yes	No	Yes	Yes

Note: Overview of risk of bias ratings for the included cross‐sectional studies, assessed with the Joanna Briggs Institute critical appraisal checklist for analytical cross‐sectional studies. Each question from the checklist was scored as “Yes,” “No,” or “Unclear” according to the study's reported methodology.

Exposure to risk factors was evaluated with standardized questionnaires and interviews, including the CBCL (Child Behavior Checklist) and the K‐SADS‐PL (Kiddie‐ Schedule for Affective Disorders and Schizophrenia for school‐age children‐ Present and Lifetime), both validated instruments with adequate internal consistency (Cronbach's α > 0.70). ²⁵ , ²⁶ , ²⁷ Some questionnaires had poor reliability, and in one study the ASD diagnosis (as a risk factor) was not established with a gold standard (e.g., autism diagnostic interview‐revised [ADI‐R]), but with the mini international neuropsychiatric interview for children and adolescents (MINI‐KID), in combination with direct observation, interviews and developmental history. ²⁸ , ²⁹

3.1.7. Limitations

All studies acknowledged limitations of their study, most commonly small sample sizes. Other reported limitations included differences in sample characteristics, the use of non‐gold standard instruments for diagnosing ASD, reliance on self‐report, and failure to account for possible confounders such as feeding disorders in first‐degree relatives.

3.2. Risk factors

Across the included studies, 10 factors were found to be significantly associated with ARFID (p < 0.05). These were grouped into four categories: physical, psychological, genetic/sex‐related and parental/familial factors (Table 3).

Table 3.

Summary of the reported associations between potential risk factors and ARFID in included studies.

Author	Associated factor(s)	Results
Canas et al. Dovey et al. Ye et al. Dovey et al.	Male sex Gestation period Birth weight Number of children in family Age of parents at birth	Male percentage in ARFID was 60.6% versus 54.5% in control group, OR 1.28 (CI 0.48–3.41) Male percentage in ARFID was 69% versus 51.76% in control group, OR 2.1 (CI 0.91–4.72) Male percentage in ARFID was 49.02% versus 45.45% in control group, OR 1.15 (CI 0.52–2.54) ARFID did not differ from control group (−1.83 ± 0.38 vs. −1.89 ± 0.32, p = 0.30) ARFID did not differ from control group (3.31 ± 0.69 vs. 3.44 ± 0.59, p = 0.66) ARFID did not differ from control group (1.83 ± 0.38 vs. 1.89 ± 0.32, p = not stated) ARFID did not differ from control group (37.4 ± 12.7 vs. 35.2 ± 10.7, p = 0.08)
Canas et al. Cerniglia et al. Cimino et al. Dovey et al.	Child emotional‐behavioral functioning*	ARFID averaged higher on the CBCL (total scores of 39.24 ± 24.38 vs. 17.70 ± 12.44, p < 0.01) than control group. ARFID (with I/I and PTFD subtype) had higher scores on most items screening for attention and behavioral problems of the CBCL. ARFID (all three profiles) had higher scores on CBCL DP (dysregulation profile) than control group (18.39 ± 5.67–35.43 ± 5.05 vs. 7.12 ± 2.83, p < 0.0001) ARFID scored higher on emotional problems score of the SDQ than control group (5.1 ± 2.9 vs. 2.4 ± 2.2, p < 0.001), but not on the other five subscales (conduct problems, hyperactivity, peer problems, prosocial, externalizing, and internalizing)
Canas et al. Inoue et al.	Comorbid psychiatric disorder* Presence and intensity of fears* Anxiety state* and anxiety trait Personal psychiatric/psychological antecedents* Presence and seriousness of depressive symptomatology* Autistic trait	ARFID had high rates of psychiatric comorbidity (81.8% vs. 33.3% in AN, p = 0.006, of which anxiety disorder in 59.3%) ARFID had higher scores on clinical fears and anxiety with FSSC‐R overall (131.58 ± 22.41 vs. 106.42 ± 20.15, p < 0.001) and anxiety state (STAIC scores 30.67 ± 7.72 vs. 24.15 ± 5.00, p < 0.001) and anxiety trait (STAIC scores 33.82 ± 8.16 vs. 29.48 ± 6.50, p < 0.001) than control group ARFID had higher rates of personal psychiatric/psychological antecedents than control group (48.5% vs. 6.1%, p < 0.01) ARFID had higher scores on CDI (overall scores of 10.55 ± 8.10 vs. 4.88 ± 3.70, p < 0.01) than control group ARFID had higher scores of autistic trait measured by AQC than control group (16.0 ± 6.9 vs. 15.3 ± 6.3, p = 0.60)
Inoue et al.	ASD	Prevalence of ASD in ARFID patients = 12.5% (4/32)
Canas et al.	Medical comorbidity*	ARFID had higher rates of medical history (42.4% vs. 12.1%, p = 0.003) compared to control group
Cerniglia et al. Cimino et al.	Maternal psychopathological risk*	Mothers of children with ARFID (all profiles) reported higher scores than control group in all subscales of the SCL‐90‐R Mothers of children with ARFID (all profiles) showed a higher global severity index (GSI) on the SCL‐90‐R than mothers of control group (0.60 ± 0.10–1.95 ± 0.22 vs. 0.18 ± 0.06, p < 0.001)
Cerniglia et al.	Mother‐child interaction*	ARFID had higher scores in all 4 subscales and total scores (total scores of 36.92 ± 6.41‐85.40 ± 6.16 vs. 7.62 ± 2.50, p = 0.001 – p = 0.01) on the SVIA compared to control group.
Cimino et al.	DAT1 genotype and total methylation*	Significant association between DAT1 genotype (9/x or 10/10 genotype) and children's diagnoses, χ 2 (3, N = 94) = 40.05, p < 0.01) with a large effect size (Cramer's V = 0.70). The I/I subtype was not associated with a specific genotype (82.6% 9/x genotype vs. 17.4% 10/10 genotype, St.R −1.3), the SFA subtype was associated with the 10/10 genotype (82.6% vs. 17.4%, St.R 6.8), the PTFD subtype and control group were associated with 9/x genotype (PTFD 91.3% vs. 8.7%, St. R 2.3 and control group 96% vs. 4%, St. R 3.1). Levels of DAT1 total methylation were higher in ARFID (I/I and PTFD subtypes) than control group (4.08 ± 1.18–8.29 ± 1.07 vs. 6.06 ± 0.54, p < 0.001)
Ye et al.	Gut microbiome*	A lower Chao1index (reflecting microbe species richness) in ARFID than control group HC versus ARFID (p < 0.001) and higher Shannon and Simpson indices (reflecting microbial species diversity) in ARFID vs. controls (p = 0.009 and p = 0.02). The abundance of gut ermF was higher in ARFID than control group (p = 0.041)
Dovey et al.	Non‐social sensory hypersensitivity* Social sensory hypersensitivity	ARFID scored higher on non‐social sensitivity on the SEQ than control group (2.8 ± 0.8 vs. 1.7 ± 0.7, p < 0.001) ARFID scores on social sensitivity on the SEQ did not significantly differ from control group (2.1 ± 0.8 vs. 1.7 ± 0.7, p = 0.06) on the SEQ

Note: Overview of main findings from studies investigating potential risk factors for ARFID in children. The table summarizes the examined factors, corresponding results, and reported group differences or statistical associations between ARFID and comparison groups. Results are presented as reported in the original studies. Asterisk (*) indicates domains where statistically significant group differences were reported.

Abbreviations: AN, anorexia nervosa; AQC, autism spectrum quotient children's version; ARFID, avoidant/restrictive food intake disorder; ASD, autism spectrum disorder; CBCL, child behavior check‐list; CDI, children's depression inventory; CI, confidence intervals; DP, dysregulation profile; FSSC‐R, fear survey schedule for children‐revised; GSI, global severity index; I/I, irritable/impulsive; NCG, non‐clinical group; OR, odds ratio; PTFD, posttraumatic feeding disorder; SCL‐90‐R, symptom checklist‐90‐revised; SDQ, strengths and difficulties questionnaire; SEQ, sensory experiences questionnaire; St.R, standardized adjusted residuals; STAIC, state‐trait anxiety inventory for children; SVIA, scala di valutazione dell'interazione alimentare.

3.2.1. Physical risk factors

Physical factors examined included medical history, gut microbiome, and perinatal characteristics, with mixed evidence for their association with ARFID. Children with ARFID had significantly more prior medical diagnoses than controls (42.4% vs. 12.1%, p = 0.003). ¹⁹ One study reported lower microbial richness and higher diversity of gut microbe species in ARFID participants (p = 0.04, p < 0.001). ²⁴ The gestation period (p = 0.30) and birth weight (p = 0.66) did not significantly differ between ARFID and controls. ²²

3.2.2. Psychological risk factors

Psychological risk factors encompassed psychiatric comorbidity, emotional‐behavioral difficulties, and autistic traits. Comorbid psychiatric disorders were common in children with ARFID (81.8%), compared with anorexia nervosa and controls. ¹⁷ , ¹⁹ , ²¹ , ²³ ASD prevalence in ARFID was 12.5%, while autistic traits did not differ significantly (p = 0.60).

Children with ARFID showed more anxiety on fear survey schedule for children‐revised (FSSC‐R) (30.7 vs. 24.2, p < 0.0001) and state‐trait anxiety inventory for children (STAIC) (131.6 vs. 106.4, p < 0.0001), and elevated attention and behavioral problems using CBCL (39.24 vs. 17.70, p < 0.001) relative to controls. ¹⁷ , ¹⁸ , ¹⁹ , ²⁰ One study also found higher scores for nonsocial hypersensitivity in ARFID patients compared to controls (2.8 vs. 1.7, p = 0.60). ²²

3.2.3. Genetic risk factors and sex

One study reported an association between the DAT1 (dopamine transporter) genotype and ARFID (χ ² [3, N = 94] = 40.05; Cramer's V = 0.70). ²¹ Variations in dopamine‐related genes such as the DAT1 may affect reward processing, impulsivity, and sensory sensitivity, all of which are features associated with ARFID. The irritable/impulsive (I/I) subtype showed no specific genotype association, whereas the sensory food aversions (SFA)‐subtype was linked to the 10/10 genotype, and the posttraumatic feeding disorders (PTFD) subtype and controls to the 9/x genotype.

Male sex was not a significant predictor of ARFID (odds ratio 1.15–2.21 [95% confidence interval [CI]: 0.48–3.41). ²² , ²³ , ²⁴

3.2.4. Parental and family risk factors

Parental and family‐related risk factors focused mainly on maternal psychopathology and mother‐child interactions. Mothers of children with ARFID scored higher on the Symptom Checklist‐90‐Revised (SCL‐90‐R), indicating greater psychological distress (p < 0.001). ²⁰ , ²¹ This elevated “maternal psychopathological risk” reflects a higher burden of self‐reported symptoms as anxiety depression and somatization. One study also reported poorer quality of mother–child feeding interaction. ²⁰ Parental age (p = 0.08) and family size (p = not stated) did not differ significantly between ARFID and controls. ²²

4. DISCUSSION

This systematic review identified several factors associated with ARFID in children, which may act as potential risk or predictors of increased vulnerability. One study found that pediatric ARFID patients more often had a burdened medical history, although it was unclear which specific conditions or symptoms were involved, such as gastrointestinal issues. ¹⁹ As medical and gastrointestinal complaints are common in ARFID, these may either precede and trigger eating disturbances—especially in the “fear of aversive consequences” profile—or arise as a result of restricted intake. ³⁰

Ye et al. found a higher diversity but a lower richness in gut microbe species of ARFID patients. ²⁴ Whether this reflects a contributing cause (e.g., by gastrointestinal discomfort) or rather an effect of ARFID‐related factors such as malnutrition or a restricted and less varied diet that could alter microbiota composition, remains unclear. Altered gut microbiota composition may influence appetite and anxiety regulation through gut–brain interactions, suggesting a possible causal role. ³¹ , ³²

Psychiatric comorbidity was highly prevalent in children with ARFID, particularly anxiety disorders, consistent with previous literature. ⁵ , ¹¹ , ³³ Although the current review did not include individuals with other eating disorders such as AN as a comparison group, it is noteworthy that this appears to be particularly prevalent in ARFID. Anxiety may increase sensory sensitivity and avoidance behavior, supporting the hypervigilance‐avoidance hypothesis. ³³ , ³⁴ ASD was found in 12.5% of children with ARFID—higher than the general population (1.7%)—likely reflecting shared features. ³⁵ such as sensory hypersensitivity and restrictive behavior. ² , ¹¹ , ²² , ²³ , ²⁴ , ²⁵ , ²⁶ , ²⁷ , ²⁸ , ²⁹ , ³⁰ , ³¹ , ³² , ³³ , ³⁴ , ³⁶

Children with ARFID showed elevated attention and behavioral problems, though the underlying mechanisms remain unclear. ¹⁹ , ²⁰ , ²¹ , ²² Attention deficits might disrupt mealtimes, while impulsivity—implicated in attention deficit hyperactivity disorder (ADHD), and other eating disorders—may contribute to disordered eating. ³⁷

Genetic findings suggest a potential role of dopaminergic mechanisms. The DAT1 genotype, which influences dopamine transport, was associated with specific ARFID subtypes. ²¹ The dopaminergic system is involved in human behavior, especially in the modulation of aggressive and impulsive behavior. Variations in dopamine regulation could underlie impulsivity and altered reward processing seen in ARFID. ³ , ³⁸ , ³⁹ Two studies found greater maternal psychopathology and poorer mother‐child interactions in families of children with ARFID. ²⁰ , ²¹ Similar patterns have been described in infantile anorexia (IA), now considered an ARFID subtype. ⁴⁰ , ⁴¹ , ⁴² Maternal distress may contribute to the child's feeding difficulties, but could also result from the stress of managing the disorder. ⁴¹

Overall, this review highlights that various biological (e.g., gut microbiome diversity), psychological (e.g., psychiatric comorbidity), genetic (e.g., DAT1‐genotype), and familial (e.g., maternal psychopathological distress) factors have been reported in association with ARFID in children. However, given the cross‐sectional design of the studies, it remains unclear whether these factors precede ARFID or arise as a consequence of the disorder. Its development likely results from the interaction of multiple vulnerability factors rather than a single causal pathway.

This review has several limitations that should be acknowledged. The included studies were heterogeneous in methods and focus, and most factors were investigated in single studies only. Cross‐sectional designs preclude causal inference, and one study relied on parent‐reported diagnosis, which reduces diagnostic certainty. Although the inclusion of this study could be debated, it was retained because the diagnosis was explicitly mentioned, parents were asked about a professional assessment, all other inclusion criteria were met, and data on this topic remain scarce. Another limitations is potential selection bias in the included studies, where systematic selection bias of the participants may have influenced the characteristics of the samples and thus the generalizability of the results. Finally, by including only studies explicitly referring to ARFID, some relevant data from overlapping conditions may have been excluded.

Future research should employ longitudinal designs to clarify causal pathways and further explore the interplay of biological, psychological, genetic, and environmental mechanisms contributing to ARFID.

5. CONCLUSIONS

Although current evidence is limited by the cross‐sectional design of existing studies, several biological (gut microbiome alterations), psychological (comorbid psychiatric disorder), genetic (DAT1‐genotype), and environmental (maternal psychopathological distress) factors appear to be more prevalent in children with ARFID. These findings represent preliminary hypotheses that require further investigation through longitudinal and interventional studies. This review provides a foundation for future work aimed at improving early identification and guide early intervention in ARFID.

CONFLICT OF INTEREST STATEMENT

The authors declare no conflicts of interest.

Supporting information

supmat.

Supplement 1.

Supplement 2.docx.

Nowacki R, Arayess L, Kleijnen J, Vreugdenhil A, Mulkens S. Risk factors for avoidant/restrictive food intake disorder in children: a systematic review. J Pediatr Gastroenterol Nutr. 2026;82:1129‐1139. 10.1002/jpn3.70362