Summary
Depends on where you sit. A systematic review of pediatric epidemiology found community prevalence from 0.3% to 15.5%, 5% to 22.5% in specialist eating disorder services, and 32% to 64% in specialist feeding clinics (Sánchez-Cerezo et al., 2023). Two recent primary-care cohorts put it at 6.4% of Dutch children under ten (Sader et al., 2023) and 5.9% of Swedish preschoolers (Dinkler et al., 2025). In a general pediatric nutrition clinic, 70% of referred children avoided at least one food group, and selectivity did not track with BMI (Gargiullo et al., 2026). The practical implication: screen on intake, not on weight.
ARFID (F50.82 in ICD-10-CM) requires a disturbance in eating or feeding, manifesting as persistent failure to meet nutritional or energy needs, with at least one of four consequences:
And the exclusions: the disturbance is not better explained by lack of available food or a culturally sanctioned practice; it does not occur exclusively during anorexia or bulimia and there is no disturbance in the experience of body weight or shape; and it is not attributable to a concurrent medical condition or better explained by another mental disorder, or if it is, its severity exceeds what that condition would explain.
The DSM-5 text also describes the three presentations that structure most assessment tools: apparent lack of interest in eating, avoidance based on sensory characteristics of food, and concern about aversive consequences of eating. These are not subtypes in the diagnostic sense, and many patients show more than one.
The NIAS (Zickgraf & Ellis, 2018, Appetite) is the most widely used brief screener. Nine statements rated 0 (strongly disagree) to 5 (strongly agree), in three subscales of three items:
Subscale scores range 0 to 15. Published cutoffs for a positive screen (Burton Murray et al., 2021) are approximately 10 or higher on picky eating, 9 or higher on appetite, and 10 or higher on fear, with a positive screen on any subscale warranting follow-up. The NIAS was validated in adults and adolescents; a parent-report version has been used in pediatric samples, with the caveat that parents of young children may not distinguish fear from sensory avoidance accurately. It is free to use and reproduce for clinical purposes.
What it is good for: intake triage, tracking subscale change over treatment, and giving families language for which presentation they are seeing. What it is not: diagnostic. A high picky eating score with normal growth, normal labs, and no functional impact is not ARFID.
The Pica, ARFID, and Rumination Disorder Interview (Bryant-Waugh et al., 2019, International Journal of Eating Disorders) is a semi-structured clinical interview with parent and child versions across age bands. It walks through each DSM-5 criterion, documents the four consequence domains with specifics, and rates severity on the sensory, lack of interest, and fear of consequences profiles. Administration is roughly 40 to 60 minutes for a full ARFID assessment.
The PARDI-AR-Q is the self-report and parent-report questionnaire version covering the ARFID section only, suited to clinics that cannot allocate interview time at intake. Both are available from the authors for clinical use.
Use the PARDI when you need to document diagnosis for a treatment plan, for insurance, or for a multidisciplinary referral. Its profile scores also tell you where to start: a high fear profile points toward CBT-based treatment; a high sensory profile toward graded exposure and food chaining with an SLP or OT; a high lack-of-interest profile toward appetite and scheduling work with a dietitian.
A parent-report screener for preschoolers, validated in Swedish primary care at routine 2.5- and 4-year check-ups (Dinkler et al., 2025). Against diagnostic interview it showed 91% sensitivity, 79% specificity, 72% positive predictive value, and 94% negative predictive value. The same group's earlier parent-report screener in a Japanese birth cohort (Dinkler et al., 2022) showed convergent validity with picky eating and satiety responsiveness measures. If you see under-fives, this is the tool; the NIAS was built for older children and adults.
The Eating Disorders in Youth Questionnaire (Kurz et al., 2015) is a brief self-report screen for 8 to 13 year olds covering ARFID presentations plus pica and rumination. Useful in school and primary care settings.
The EDE and EDE-Q were built for weight and shape concern and underperform for ARFID. An ARFID module for the EDE exists (Schmidt et al.) for research settings. In clinic, the PARDI replaces it.
For children under five or any child with a suspected oral motor or sensory component, an SLP or OT feeding evaluation (clinical feeding assessment, and where indicated a modified barium swallow or FEES) comes before or alongside the ARFID assessment. The Infant and Child Feeding Questionnaire and the Behavioral Pediatrics Feeding Assessment Scale (BPFAS) quantify mealtime behavior and parent concern.
Not a validated instrument, but the single most useful document in an ARFID workup: every food the child reliably accepts, with brand and preparation, grouped by food group. Under 20 items, a missing food group, or a shrinking list are each informative. Ask the family to bring it, or have them keep it in an app so it is current at every visit.
The consequence criterion is established with data, not impression. At minimum:
The dividing line is body image. Ask directly about fear of weight gain, wish to be thinner, and body checking, and ask it more than once in teens, who may initially deny it. Weight loss that the patient is distressed about supports ARFID. Note that ARFID can precede or transition into anorexia, and the two can be hard to separate in a 13-year-old with sudden restriction.
PFD (Goday et al., 2019; ICD-10-CM R63.3x) covers impaired oral intake not appropriate for age, lasting at least two weeks, with dysfunction in at least one of four domains: medical, nutritional, feeding skill, psychosocial. It is the better fit when there is a clear medical or skill driver, and in most children under two. ARFID and PFD can co-occur; many feeding teams use PFD as the organizing diagnosis and add ARFID where the psychiatric criteria are met and the child is old enough for the avoidance to be considered volitional.
Eosinophilic esophagitis, celiac disease, inflammatory bowel disease, gastroesophageal reflux, food allergy, dysphagia, and constipation can all present as food refusal, and several can also trigger fear-based ARFID that outlasts the medical cause. Treat the medical condition and reassess.
Neither excludes ARFID. A 2025 meta-analysis of 21 studies found autism in 16.3% of ARFID samples and ARFID in 11.4% of autistic samples (Sader et al., 2025); sensory-based avoidance is the most frequently described driver in autistic children (Bourne et al., 2022). Children with ADHD show more ARFID-related selective eating and more oral and multisensory processing difficulty than peers, with selective eating tracking ADHD symptom severity (Kaşak et al., 2025). Diagnose ARFID when criteria are met; it changes treatment access and intensity. Assessment pacing should respect the child's need for predictability.
Allergy is both a differential and a risk factor. In a prospective controlled study of 234 children aged 2 to 9, those with food allergy scored higher on NIAS-PR picky-eating and fear subscales than matched controls, with fear highest after urticaria or anaphylaxis (Karaca Şahin et al., 2026). Screen allergy patients for fear-of-consequences ARFID, especially after an immediate-type reaction.
Peaks at two to six, list grows slowly, growth normal, labs normal, no functional impact. Reassure, teach responsive feeding, recheck growth in three to six months. The parent-facing guide to ARFID vs picky eating is written to help families make this distinction before they reach you.
A note on what treatment evidence exists. A 2024 scoping review found the psychological intervention literature dominated by case reports and pilots, with CBT-AR and family-based approaches most promising (Willmott et al., 2024); no pharmacological agent has RCT evidence for ARFID itself (Archibald & Bryant-Waugh, 2023). The first sizeable RCT, from Stanford in 2026, randomized 98 underweight 6 to 12 year olds to FBT-ARFID or psychoeducational motivational therapy; both reduced ARFID symptoms, FBT-ARFID produced significantly greater weight gain, and higher baseline severity favored FBT (NCT04450771). Treatment selection is still clinical judgment informed by presentation profile.
Assessment does not end at diagnosis. Treatment response in ARFID is measured in foods accepted, exposure steps completed, growth trajectory, and change in the fear and interest profiles over months. All of that happens at home, and the quality of the data you get back decides how well you can adjust the plan.
Heartful Sprout's clinician platform was built for this: you set the cause being treated, the food hierarchy, and each food's current exposure step; the family logs the attempt and how the child felt in the Heartful Eating Recovery app; intake is computed against DRIs from USDA data; and growth plots on WHO and CDC curves with z-scores. The record comes back to you as trends and safety flags before the next session. See how feeding therapists use it, or book a demo.
Parent-facing handouts you can assign: ARFID signs in children, ARFID safe foods, and how to help a child with ARFID at home.
This page is for clinician education and does not replace clinical judgment or local protocols.
Set the plan in the dashboard. The family carries it out at home. The record comes back to you.
See the demo