Assessment and treatment of avoidant/restrictive food intake disorder

Abstract

Purpose of Review:

To review the literature pertaining to the assessment and treatment of avoidant/restrictive food intake disorder (ARFID), ten years following its introduction to DSM-5 (APA, 2013).

Recent Findings:

Several structured clinical interviews for assessing ARFID have been developed, each with its own strengths and limitations. There is no clear leading self-report measure for tracking treatment progress and outcome in ARFID. Medical assessment is comprised of examining anthropometrics, vitamin deficiencies, and other comorbidities. To date, several studies have reported on cognitive-behavioral therapy, family-based treatment, and other approaches to the treatment of ARFID. These treatments appear promising; however, they rely on data from clinical case series and very small randomized controlled trials.

Summary:

Several promising assessments and treatments for ARFID are in the early stages of research. Yet, controversies remain. These include: (a) overlap with criteria for pediatric feeding disorder; (b) the optimal method for assessing nutrient deficiencies; (c) disciplines involved in treatment. Future research innovation is necessary to improve the psychometric properties of ARFID assessments and evaluate treatment efficacy with larger samples and randomized designs.

Introduction

Avoidant/restrictive food intake disorder (ARFID) was first introduced in the fifth edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM-5¹) as an expansion and reformulation of the DSM-IV² diagnosis of feeding disorder of infancy and early childhood. ARFID is characterized by avoidant/restrictive eating by volume (i.e., restriction of amount of food) and/or variety (i.e., avoidance of specific food) resulting in weight loss or faltering growth, nutritional deficiencies, dependence on enteral/supplemental feeding (i.e., any enteral feeding not required by a concurrent medical condition or ≥ 50% of daily caloric intake via oral supplementation³) and/or interference with psychosocial functioning⁴. DSM-5-TR⁴ provides examples of three ARFID profiles: (a) sensory sensitivity; (b) fear of aversive consequences; and (c) lack of interest in food/eating. In contrast to previous categorical models of feeding disorders, Thomas and colleagues’⁶ three-dimensional neurobiological model of ARFID posits that the three core ARFID presentations occur along a continuum of severity and are not mutually exclusive. This model is supported by growing evidence suggesting that an individual with ARFID may present with difficulty across one, two, or all three domains, ranging in severity^7,8,9.

History of the ARFID Diagnosis

DSM-5 criteria for ARFID¹ introduced ambiguity regarding whether psychosocial impairment alone (i.e., in the absence of weight loss, nutritional deficiencies, and supplement dependence) could constitute an ARFID diagnosis³. A recent study that examined a more inclusive definition of ARFID where psychosocial impairment was a sufficient criterion for diagnosis found that the sample was, on average, within a normal weight range⁵. Fortunately, DSM-5-TR⁴ clarified this ambiguity by explicitly expanding the diagnostic criteria to encompass individuals who experience only interference with psychosocial functioning as a result of their avoidant/restrictive eating.

ARFID’s induction in DSM-5 filled an important clinical gap; namely, the need for a feeding and eating disorder characterized by avoidant/restrictive eating, without shape and weight disturbance, that spans the lifespan and weight spectrum. The diagnosis provided a nosological home for individuals who did not fit the criteria for feeding disorder of infancy and early childhood and decreased the number of cases that could only be classified as eating disorder not otherwise specified (EDNOS¹⁰; which previously comprised 40–60% of eating-disorder cases under DSM-IV¹¹). In 2019, however, Goday and colleagues¹² proposed new diagnostic criteria for pediatric feeding disorder which overlapped substantially with criteria for ARFID, such that many individuals with avoidant/restrictive eating could meet criteria for both. One of the reasons underlying the new pediatric feeding disorder criteria was concern that ARFID “requires that severity of the eating disturbance exceeds that associated with the [concurrent medical] condition and specifically excludes children whose primary challenge is a skill deficit”¹² (p. 125). Such medical conditions may include disorders that affect oral/nasal/pharyngeal function, aerodigestive diseases, and congenital and other heart diseases¹². For instance, an individual with an aerodigestive disease may meet criteria for pediatric feeding disorder, but not ARFID, because feeding difficulties are attributed to the aerodigestive disease and neither exceed that routinely associated with this condition nor warrant additional clinical attention. These new criteria for pediatric feeding disorder highlighted the need to consider both context of avoidant/restrictive eating and developmental stage³.

Distinguishing Between ARFID and Other Eating Disorders

The core feature that distinguishes ARFID from other eating disorders—such as anorexia nervosa (AN) and bulimia nervosa (BN)—is “no evidence of a disturbance in the way in which one’s body weight or shape is experienced”⁴ (p. 376). For this reason, ARFID cannot be concurrently diagnosed with AN or BN. Binge-eating disorder (BED), however, can occur with or without weight/shape concerns, and ARFID and BED can co-occur⁴. Of note, “no evidence of a disturbance in the way in which one’s body weight or shape is experienced”⁴ (p. 376) may be a high threshold for diagnosis. Individuals with ARFID are not immune to societal influences of shape/weight and, even individuals without eating disorders express body image concerns^13,14. Of note, individuals with ARFID in some studies^15,16 reported behaviors characteristic of other eating disorders (e.g., binge eating, self-induced vomiting, excessive exercise), and still others⁵ have developed de novo shape/weight concerns concurrent with their ARFID diagnosis during treatment. Given preliminary evidence of ARFID symptoms co-occurring with other eating disorder symptoms, caution should be exercised when evaluating the body image disturbance criterion and clinical judgment is necessary to monitor and document any changes to ARFID clinical presentation.

Assessment

When ARFID was added to the psychiatric nomenclature with the 2013 publication of DSM-5, there were no available structured clinical interviews or self-report measures to assess this diagnostic construct. Over the past ten years, assessments have been developed for the purposes of screening, evaluation, and assessing treatment outcome. Table 1 compares and contrasts the features of the four structured interviews and four self-report measures that have recently been developed (or adapted) for the assessment of ARFID.

Table 1:

Assessments for ARFID: Important considerations for clinicians and researchers when identifying which assessment to use

	Brief	ARFID profiles	ARFID severity	Validated for use in ARFID	Differential diagnosis with other EDs	Sensitivity to change pre-post treatment of ARFID
Structured Clinical Interviews
EDA-5 Sysko et al., 2015)	X				X
SCID-5 (First et al., 2014)					X
EDE – ARFID Module (Schmidt et al., 2019)	X	X		X	X
PARDI (Bryant-Waugh et al., 2019)		X	X	X		X
Self-Report Questionnaires
FNS (Pliner & Hobden, 1992)	X					X
NIAS (Zickgraf & Ellis, 2018)	X	X		X		X
PARDI-AR-Q (Bryant-Waugh et al., under revision)	X	X	X	X		X
EDY-Q (Kurtz et al., 2015)	X	X		X	X

Note: EDA-5 – Eating Disorder Assessment for DSM-5; SCID-5 – Structured Clinical Interview for DSM-5; EDE – Eating Disorder Examination; PARDI – Pica, ARFID, and Rumination Disorder Interview; EDE-Q – Eating Disorder Examination – Questionnaire; FNS – Food Neophobia Scale; NIAS – Nine Item ARFID Screen; PARDI-AR-Q – Pica, ARFID, and Rumination Disorder – ARFID Questionnaire.

Psychological Evaluation

Structured Clinical Interviews.

The Eating Disorder Assessment for DSM-5 (EDA-5¹⁷), Structured Clinical Interview for DSM-5 (SCID-5¹⁸), ARFID module of the Eating Disorder Examination (EDE¹⁹), and Pica, ARFID, and Rumination Disorder Interview (PARDI²⁰) have all recently been developed to assess ARFID. The EDA-5¹⁷ is a brief, online semi-structured clinical interview developed to determine DSM-5 feeding and eating disorder diagnoses. The EDA-5 has extensive validity data for other eating disorders (e.g., anorexia nervosa, bulimia nervosa), but less data on ARFID. The EDA-5 relies on an algorithm that includes questions based on respondents’ prior answers. Therefore, the time it takes to administer may vary between individuals. The SCID-5¹⁸ is a semi-structured interview used to confer DSM-5 diagnoses. The Feeding and Eating Disorders Module assesses diagnostic criteria and age of onset for all feeding and eating disorders. The ARFID module of the EDE has three versions – an adult version, a child version, and a parent version. The module takes ~20 minutes to administer and can serve as either a stand-alone assessment or as an addition to the overall EDE (which assesses already assesses for AN, BN, and BED). The ARFID module begins by assessing presence of avoidant/restrictive food intake over the prior three months, with subsequent items assessing diagnostic criteria. The final section of the module includes questions pertaining to the ARFID profiles. Finally, the PARDI²⁰ is a semi-structured interview that can be used to assess ARFID in children and adults. This assesses both the presence and severity of ARFID and provides dimensional scores for each of the three ARFID profiles. The PARDI also assesses constructs central to ARFID psychopathology (e.g., dietary variety, food neophobia, psychosocial impairment). The PARDI is appropriate for use with children, adolescents, and adults. There are four versions available: parent/carer (2–3 years), parent/carer (ages 4+ years), child (8–13 years), and young person/adult (14+ years).

There are pros and cons to each interview. For example, only the ARFID module of the EDE and the PARDI have been validated for use with ARFID; the EDA-5 and SCID-5 both require validation studies using an ARFID sample. The EDA-5 provides a quick (i.e., 10–15 minute) assessment of presence/absence of ARFID using a freely available mobile “app” format but does not assess ARFID severity or profiles. The SCID-5 is more comprehensive than the EDA-5 but has the same limitations. This module can be used to diagnose ARFID and provides an assessment of each profile. Finally, the PARDI provides the most detailed clinical picture, evaluating both the presence of ARFID and overall severity and match to each of the profiles. Due to its comprehensiveness, the PARDI is the longest clinical interview (~45 minutes) posing a barrier to its practical use in treatment settings. Importantly, in addition to assessing ARFID, all four of these structured clinical interviews assess (to varying degrees) for the presence of other feeding and eating disorders, which is important for differential diagnosis and treatment planning.

Self-Report Measures.

The Food Neophobia Scale (FNS²¹), Nine-Item ARFID Screen (NIAS²²), self-report version of the PARDI (PARDI-AR-Q²³ and Eating Disturbances in Youth – Questionnaire (EDY-Q²⁴) are self-report measures that may be used to assess the specific psychopathology of ARFID. There is no clear leading self-report measure for tracking treatment progress and outcome in ARFID.

Measures of eating disorder psychopathology, such as the Eating Disorders Examination–Questionnaire (EDE-Q²⁵) – a self-report measure derived from the EDE – are better conceptualized as a method for assessing shape/weight concerns (e.g., AN, BN) that would rule out a diagnosis of ARFID³¹. Moreover, though the Clinical Impairment Assessment (CIA²⁶) assesses psychosocial impairment resulting from other eating disorders, it does not capture impairment related to ARFID. The CIA question stem instructs respondents to indicate how their eating habits, exercising, or feelings about their eating, shape, or weight have affected their life over the prior four weeks²⁶. While “eating habits” may reasonably refer to avoidant/restrictive eating consequent to ARFID, the lack of specificity, coupled with the reference to exercising and shape/weight may preclude those with ARFID from endorsing impairment. Indeed, most individuals score below the clinical cutoff of 16 for psychosocial impairment^5,27. To date, no stand-alone measures exist that specifically capture ARFID-related impairment. Such measures are especially important to develop for collateral reporters of younger individuals who may have low insight and the impairment burden is borne by parents/caregivers.

The FNS is a 10-item measure that measures reluctance to try new foods (e.g., “I do not trust new foods,” “I am afraid to eat things I have never had before”). Though the FNS was originally validated for use in adults ≥ 18 years, it has been used in children as young as seven years²⁸, and has been adapted in a child version²⁹. Higher scores on the FNS are indicative of behavioral responses to novel foods (e.g., eating fewer unfamiliar foods presented). The NIAS was initially developed as a screening measure for ARFID symptoms. It contains nine items that map onto three subscales that assess the three ARFID profiles (e.g., sensory sensitivity – “I dislike most foods that other people eat,” fear of aversive consequences “I avoid or put of eating because I am afraid of GI discomfort, choking, or vomiting,” and lack of interest – “Even when I am eating foods I really like, it is hard for me to eat a large enough volume at meals”). Since its inception, it has been translated to multiple languages³⁰ and validated for use in ARFID. Responses are rated on a 0 (Strongly disagree) to 5 (Strongly agree) scale and higher scores on each subscale indicate higher levels of each profile. The NIAS may be particularly useful for collecting supplemental information pertaining to profiles if there is not sufficient time to administer the PARDI. Burton Murray and colleagues³¹ recently validated the NIAS subscales for distinguishing ARFID presentations and screening for ARFID. These authors proposed cutoff scores for each profile (i.e., ≥ 10 for the picky eating and fear subscales to distinguish sensory sensitivity and fear of aversive consequences, respectively, and ≥ 9 for the appetite subscale, used to assess the lack of interest in food/eating profile) that can be used to quickly assess match to each ARFID profile based on self-report. Both FNS and NIAS have shown sensitivity to change pre- to post- ARFID treatment, making them potentially useful measures to evaluate treatment progress and outcome^27,22,23. Unfortunately, neither measure has been validated against clinical interview derived ARFID diagnosis.

The PARDI-AR-Q²³ is a newer measure that evaluates the presence and severity of ARFID and each of the three profiles. The PARDI-AR-Q is a 32-item self-report measure that has been validated against interview-derived ARFID diagnosis and demonstrates good construct, convergent, and discriminant validity. It begins with demographic and anthropometric questions. It then includes 12 dichotomous questions that assess ARFID diagnostic criteria (e.g., “Has any health professional said that you have a nutritional deficiency due to your eating habits?”), some of which are followed up with free-text follow-up questions. These items are used to calculate a positive or negative ARFID screen, where a positive screen suggests that a follow-up clinical interview is necessary to evaluate the presence of ARFID. Next, 11 Likert-type items ranging from 0 (Never) to 6 (Always) measure the presence and severity of each profile (e.g., “Please indicate a number below regarding how often you have forgotten to eat or found it difficult to make time to eat, ranging from 0 (Never) to 6 (Always)”). The PARDI-AR-Q is the only self-report ARFID measure that assesses psychosocial impairment.

Finally, the EDY-Q²⁴ was developed to identify early-onset restrictive eating disturbances in children 8–13 years. The EDY-Q consists of 14 items, including one question on pica and one question on rumination disorder. The remaining 12 items cover ARFID, including two questions that assess shape/weight concerns as exclusion criteria for diagnosis. There are six diagnostic items: three items assess the three ARFID profiles (i.e., sensory sensitivity, fear of aversive consequences, lack of interest), one item assesses difficulty meeting adequate weight growth (underweight) and two items asses shape/weight concern. Items are rated on a 7-point Likert scale ranging from 0 (Never) to 6 (Always). Cut-off scores for the EDY-Q diagnostic items have been identified, such that to meet criteria for ARFID, a child must report a score ≥ 4 on the diagnostic items and a score < 3 on the shape/weight concern exclusionary items.

In sum, four self-report measures for ARFID symptoms exist, each with its own strengths and limitations. Coupled with a structured clinical interview, administration of self-report measures provides a comprehensive clinical picture and aid in tracking treatment outcomes.

Medical Evaluation

A comprehensive medical evaluation is critical prior to initiating treatment³⁴. ARFID is associated with a host of medical sequalae including nutritional deficiencies, electrolyte abnormalities, bradycardia, gastrointestinal complications, amenorrhea, and bone loss. A medical provider (e.g., physician, nurse practitioner) can evaluate whether a patient is an appropriate fit for outpatient treatment or whether they would benefit from additional support (e.g., enteral feeding) provided at a higher level of care (e.g., inpatient hospitalization).

Anthropometrics

Assessment of a patient’s growth history and the need for weight gain (for all patients) or height growth (for children and adolescents) is critical to set appropriate weight targets for patients who are underweight. Many medical sequelae of ARFID (e.g., amenorrhea) are associated with low weight. For patients who are underweight, the medical provider first determines a target weight. For underweight patients with an acute onset of illness, growth charts that can be used to return the patient to their premorbid weight trajectory³⁵. For patients with ARFID who are chronically underweight, determining a target weight may be more challenging. For both children and adults, Brigham and colleagues³⁴ recommend aiming for a target weight high enough to allow the patient to gain the height expected for their sex and age. For children and adolescents, the target weight should allow patients to progress through puberty appropriately and will increase over time given expected increases in height and BMI.

Nutritional Deficiencies

Nutritional deficiencies are common in ARFID, and if left untreated, can lead to a host of negative consequences. For instance, patients who avoid animal products may be at risk for B12, zinc, iron, or protein deficiencies. Each of these deficiencies may manifest in signs and symptoms³⁶. Consequences of a B12 deficiency may contribute to low energy, poor growth, dry skin, hair loss, anemia, and eye cataracts. A zinc deficiency may lead to poor growth and development, taste and smell changes, and weakened immune function, while low iron can contribute to difficulty thinking clearly and attention problems, low energy, irritability, trouble maintaining body temperature, and a weakened immune system. Finally, a protein deficiency can contribute to loss of lean body mass and decreased energy. Other common deficiencies include calcium, folate, Vitamins A, B, C, D and K, with themselves are associated with a host of sequalae, undermining the importance of assessing deficiencies among patients with ARFID. Importantly, some labs can be abnormal (e.g., slightly low Vitamin D) in the absence of troublesome symptoms or may be abnormal even in healthy individuals. Thus, the clinical significance of these labs may be unclear.

Expert perspectives differ on how to assess nutritional deficiencies in ARFID. Though some providers believe that assessment of intake via self-report dietary recall is sufficient to estimate deficiencies, others argue for the use of blood work³. Completion of both (i.e., self-report dietary recall and blood work) provides the most comprehensive picture. Individuals taking multivitamins/minerals preventatively or relying on vitamin-fortified foods may be correcting for nutrients absent in their diet. Though this would not necessarily be identified by blood work, a provider may identify such gaps through dietary recall, highlighting the importance of both objective and subjective reports. Regardless of the way in which it is assessed, information about nutrient deficiencies is critical for treatment planning as it helps inform adequacy of dietary variety and caloric needs necessary to maintain growth and development. Further, nutrient deficiencies are a critical treatment target that could be supplemented directly with pills or drops or provided in novel foods introduced in treatment.

Comorbidities

Other components of the medical assessment include obtaining a list of current medications and assessing any other medical comorbidities that may impact feeding. For instance, food allergies are particularly important to assess, especially for patients with the fear of aversive consequences profile who may fear having an allergic reaction. Results of food allergy testing help inform whether the feared food can be re-introduced during in-session exposures. Other important medical comorbidities include oral-motor difficulties and gastrointestinal problems (e.g., celiac, Crohn’s disease). Brigham and colleagues³⁴ further outline important components of a medical evaluation for ARFID that aid with the assessment of associated features and sequelae of avoidant/restrictive eating. These might include administering an electrocardiogram to a patient with bradycardia or hemodynamic instability, checking human chorionic gonadotropin (HC) in post-menarchal females with amenorrhea, and assessing bone density using dual-energy X-ray absorptiometry (DXA) in patients who have menstruated fewer than six times in the past year^34,37. Finally, throughout the comprehensive evaluation (i.e., both psychological and medical) collateral reports (e.g., from primary caregivers) are especially important given ARFID’s younger age at presentation compared to other eating disorders that onset later in life^5,38,39.

Summary

Together, a comprehensive evaluation consisting of both a psychological and medical component help guide treatment selection and planning. To date, most research progress has been made in the area of psychological evaluation, with several clinical interviews and self-report measures that are useful for ascertaining the presence and severity of ARFID. There are pros and cons dependent upon desired use for each (Table 1). Most self-report questionnaires, for example, need to be used in conjunction with another measure of other eating disorder psychopathology. Individuals with eating disorders characterized by overvaluation of shape and weight (e.g., AN, BN) may produce similar scores to those with ARFID³¹ on ARFID measures, necessitating additional self-report assessments to aid with differential diagnosis.

Myriad reasons contribute to the necessity of a medical assessment for patients with ARFID. These include guiding treatment selection (i.e., level of care) and properly ascertaining each element of Criterion A for diagnosis. Caution is warranted when determining a target weight, as this determination may be challenging for chronically underweight patients with ARFID, and for young patients whose target weight will correspond to changes in height³⁴. Nutrition deficiencies are important to assess because they lead to a host of consequence and comprise an important treatment target (e.g., introduction of foods high in deficient nutrients). Medications and medical comorbidities are also critical to assess. Coupled together, the psychological and medical evaluation provide a comprehensive overview of the patient’s ARFID presentation and treatment needs.

Treatment

Several studies have reported on patient outcomes following treatment for ARFID. Most of these treatments are based upon approaches for treating other eating disorders and rely primarily on findings derived from small sample sizes. Such studies have included case reports and small case series^{40,41,42,,44,45,54}, retrospective chart reviews^43,46,47,48, medication case series^46,49, and treatments of conditions similar to ARFID (e.g., pediatric feeding disorder⁵⁰ and food avoidance⁵¹). ARFID-specific treatments are in the early stages of research, and no studies to date have implemented fully powered randomized controlled trials to evaluate treatment efficacy. Nonetheless, preliminary outcomes appear promising. Here, we focus on studies that specifically examined individuals diagnosed with ARFID, using sample sizes > 10. Three studies^{27,33, 52} utilized cognitive behavioral therapy (CBT), one⁵³ used family-based therapy (FBT), and two^55,61 implemented an intensive multidisciplinary intervention. Table 2 provides an overview of these studies, specifying treatment and level of care, duration, sample size, age, sex, family involvement, disciplines, and treatment target.

Table 2.

Overview of treatments for ARFID

Authors	Treatment (LOC)	Duration	N	Age (years)	Sex (%female)	Family Involvement	Disciplines	Treatment Targets
Dumont et al. (2018)	Exposure-based CBT (Partial)	4-week day treatment	11	10–17	36	Yes	Mental health clinician	Food neophobia, food acceptance, food selectivity, weight gain
Lock et al. (2019b)	FBT-ARFID (Outpatient)	Up to 22 sessions over 6 months	16	5–12	44	Yes	Mental health clinician (with concurrent medical monitoring)	ARFID symptom severity, weight gain, decreased symptoms of depression/anxiety, parental-self efficacy
Sharp et al. (2016)	IMI (Day Program)	5-day day treatment, consisting of 14, 40-minute meal blocks	20	1–6	40	Yes	Psychologist, registered dietician, speech language pathologist, occupational therapist, social worker, nurses, pediatric gastroenterologist	Bite acceptance, mealtime disruptions, volume of food consumed, weight gain
Thomas et al. (2020)	CBT-AR (Outpatient)	20 weekly sessions, up to 30 if underweight	20	10–17	55	Depends on age/weight status (no for ≥ 16 years and non-underweight)	Mental health clinician (with concurrent medical monitoring)	Dietary diversity, ARFID symptom severity, food neophobia, weight gain, psychosocial impairment
Thomas et al. (2021)	CBT-AR (Outpatient)	20 weekly sessions, up to 30 if underweight	15	18–55	67	No	Mental health clinician (with concurrent medical monitoring)	Dietary diversity, ARFID symptom severity, food neophobia, weight gain, psychosocial impairment
Volkert et al. (2021)	IMI (Inpatient)	Average (SD) admission = 36 (7) days	AS	2–13	17	Yes	Psychologist, physician, nurse practitioner, registered dietician, speech language pathologist, occupational therapist, social worker	Introduction of a nutritionally complete diet, restructuring parent-child interactions during meals

Note: CBT-AR – Cognitive Behavioral Therapy for Avoidant/Restrictive Food Intake Disorder; FBT – family-based therapy; LOC – level of care; ARFID – avoidant/restrictive food intake disorder; IMI – intensive multidisciplinary intervention; SD – standard deviation.

Cognitive-Behavioral Therapy

Cognitive-behavioral therapy for ARFID (CBT-AR)³⁶ is a specialized treatment suitable for all ARFID profiles (i.e., sensory sensitivity, fear of aversive consequences, and lack of interest in food/eating) and demonstrates evidence of feasibility, acceptability, and proof-of-concept for children, adolescents, and adults (ages 10 and up). It draws on principles from first-line CBT treatments for other eating disorders⁵⁶, lending further credibility to its use with ARFID. CBT-AR is a flexible and modular treatment currently implemented worldwide. Treatment spans 20–30 sessions depending on a patient’s weight status (i.e., 30 sessions for patients who have significant weight to gain) over the course of 6–12 months. Given that it is an outpatient treatment, CBT-AR is appropriate for patients who are medically stable, currently accepting at least some food by mouth, and not receiving enteral feeding. Both individual and family-supported versions exist, the latter used for children and adolescents (10–15 years) and young adults (≥ 16 years) who live at home and have significant weight to gain. The key difference between family-supported and the individual versions of CBT-AR is who participates in sessions – the interventions are the same. In family-supported CBT-AR, the patient’s primary caregivers attend all sessions with the patient and hold the primarily responsibility for implementing change. In individual CBT-AR, the patient attends sessions alone and holds primary responsibility.

CBT-AR spans four stages, each focusing on a different treatment target. Stage 1, which typically spans 2–4 sessions, is oriented toward providing psychoeducation on ARFID and CBT-AR, stressing the importance of early change. Here, patients (or parents, in the case of children) begin self-monitoring and establishing a regular pattern of eating (i.e., three meals and 2–3 snacks, typically dependent upon weight status) relying on their preferred foods (that is, exposure to novel foods does not begin until later on in treatment), increasing volume (for underweight patients) and/or variety of preferred foods, and creating the individualized formulation of the patient’s maintaining mechanisms (i.e., sensory sensitivity, fear of aversive consequences, lack of interest in food/eating). In Stage 2 (spans two sessions), treatment planning begins. The mental health clinician provides psychoeducation about the five basic food groups and nutritional deficiencies, and the patient selects new foods to begin learning about in Stage 3. Stage 3, the heart of treatment, spans anywhere from 14–22 sessions and addresses maintaining mechanisms in each ARFID domain by utilizing exposure principles. For sensory sensitivity, exposure involves systematic desensitization to novel foods; for fear of aversive consequences, development of a fear/avoidance hierarchy and graded exposure to that hierarchy; and for lack of interest in food and/or eating, interoceptive exposures and in-session exposures to preferred foods. Finally, Stage 4 (spans two sessions) focuses on relapse prevention, evaluating whether the patient’s treatment goals have been met and identifying CBT-AR strategies to continue implementing at home.

Thomas and colleagues have reported on outcomes of CBT-AR for children and adolescents³³ and adults²⁷. In the child/adolescent sample³³, mental health clinicians rated 85% of patients as ‘much improved’ or ‘very much improved’ by end of treatment. Dietary diversity increased, with patients incorporating an average of 17 novel foods into their diets. Underweight patients gained an average of 11 lbs, progressing from the 10^th to the 20^th percentile for BMI. Finally, by end of treatment, 70% of patients no longer met diagnostic criteria for ARFID. A similar pattern of findings emerged from an adult sample²⁷. In this study, patients highly rated treatment credibility and anticipated improvement, and 93% of patients who completed treatment provided high satisfaction ratings. Dietary diversity increased; patients incorporated an average of 18 novel foods into their diet. Underweight patients gained weight, and by end of treatment 47% of patients no longer met diagnostic criteria for ARFID. Of note, these therapeutic gains occurred in the absence of a large multidisciplinary team. The patients were treated by only a mental health clinician and medically monitored by a physician.

Another study implemented a 4-week exposure-based CBT⁵², demonstrating that 11 patients with ARFID were able to increase their dietary variety, decrease food neophobia, and improve their psychosocial functioning. At 3-month follow-up, 91% of patients in this study were in remission from ARFID with a healthy body weight. Therefore, CBT appears to have evidence of feasibility, acceptability, and preliminary efficacy for ARFID, though randomized controlled trials are needed.

Family-Based Treatment

FBT for ARFID (FBT-ARFID⁵⁷) is a manualized treatment based on FBT for adolescents with AN⁵⁸ and BN^59. FBT-ARFID is an outpatient treatment consisting of three phases. Treatment duration depends on a patient’s presenting mechanism; it typically spans 4–5 months for the sensory sensitivity and lack of interest profiles, and between 3–5 months for fear of aversive consequences, given that the acute onset of this profile typically necessitates a briefer course of treatment. Interventions central to FBT-ARFID are the same as those used for AN and BN and are implemented regardless of the patient’s maintaining mechanism^53,57,60 include: (a) lack of focus on the etiology of ARFID; (b) parental empowerment; (c) promoting externalization of ARFID to reduce family self-blame and guilt; (d) emphasis on the serious medical and developmental consequences if ARFID is not addressed (this is typically achieved by orchestrating an intense scene to promote the need for immediate parental intervention); (e) the mental health clinician facilitates parental learning rather than prescribing or directing them; (f) a pragmatic, behavioral focus on changing dysfunctional eating behaviors (versus a focus on family processes or patient dynamics); and (g) lack of emphasis on cognitive distortions.

Treatment targets for FBT-ARFID vary depending upon maintaining mechanism. For those presenting with the sensory sensitivity profile, the principal treatment target is consistent implementation of dietary diversity; for fear of aversive consequences, decreased anxiety and fear related to eating is the key goal (weight gain is also often important but varies dependent upon how rapidly anxiety and fear related to eating attenuate); and for the lack of interest profile, the treatment target is weight gain and decreased mealtime duration. These interventions are implemented in three phases, which vary slightly to accommodate the different treatment targets. Regardless of the treatment target, Phase 1 encourages parents to act against dysfunctional eating behaviors by highlighting the negative impact of these behaviors on the patient, family, and developmental trajectory of the child. This is in part accomplished by a family meal that allows the mental health clinician to consult directly with the parents on their management of mealtime behaviors and initiate the behavioral change learning process. In Phase 2, treatment shifts to transitioning some of the control over eating back to the patient. Finally, in Phase 3, developmental issues are addressed. Of note, when working with young patients, both CBT-AR and FBT emphasize strong parental involvement around increasing dietary volume; however, they diverge with respect to parental involvement around expanding dietary variety. Whereas in FBT, changes to variety are parent-driven, in CBT-AR, these changes are patient-led.

In a pilot randomized feasibility trial of FBT-ARFID⁵³ for patients ages 5–12 years, FBT-ARFID was more effective than usual care (typically consisting of pediatrician visits only) in improving severity of ARFID symptoms, weight, and parental self-efficacy. Most families utilized 14 sessions over the course of four months. FBT-ARFID emphasizes parental self-efficacy, and changes in parental self-efficacy were correlated with improvements in patients’ weight status, highlighting parental self-efficacy as a potential mediator of symptom change. Both underweight and normal weight patients gained weight in FBT-ARFID, and ARFID severity (assessed using the PARDI) decreased following treatment. Taken together, these findings indicate that FBT-ARFID is a potentially efficacious treatment for childhood ARFID, worthy of further study.

Other Treatment Approaches

With respect to other approaches, intensive multidisciplinary intervention is the standard of care for patients with pediatric feeding disorders¹², many of whom would also meet diagnostic criteria for ARFID. Treatment is typically delivered at a day hospital or inpatient level of care where patients can be supported through several mealtimes per day (over a prolonged period) and overseen by a large multidisciplinary treatment team (e.g., psychologist, physician, nurse practitioner, registered dietician, speech language pathologist, occupational therapist, social worker). Treatment targets for intensive multidisciplinary intervention include improving nutritional status, dietary diversity, and mealtime behaviors (e.g., bite acceptance, mealtime disruptions) associated with food avoidance/restriction. Following intensive multidisciplinary intervention, patients with ARFID expanded their dietary variety, resulting in a more nutritionally complete diet^55,61. Patients also demonstrated significant behavioral change, such as bite acceptance and swallowing, and reduction in inappropriate mealtime behavior.

Summary

CBT, FBT, and intensive multidisciplinary intervention share many common elements. Treatment targets are similar across studies (Table 2), and typically include regular eating, weight gain (if underweight), and increasing food volume/variety. Key differences between treatments include duration of therapy, level of care, family involvement, and disciplines required for treatment. Most approaches, however, stress family involvement for younger patients and include a mental health clinician with concurrent medical monitoring.

Clinical opinions diverge on how many team members of which disciplines are required to adequately treat ARFID. In additional to mental health clinicians (e.g., psychologists, social workers), other team members with relevant expertise include dietitians, speech/language pathologists, occupational therapists, psychiatrists, gastroenterologists, and endocrinologists. Bryant-Waugh and colleagues⁶² proposed an outpatient care pathway for children and young people with ARFID that includes a multidisciplinary, multi-modal approach to the assessment of treatment of the disorder. The pathway considers indications for different types of interventions that may be necessary given a patient’s clinical presentation. It then delineates relevant treatments, including psycho-behavioral, medical, dietetic, and others, that are delivered simultaneously using a multi-modal approach. Though the pathway requires testing, it provides structure and guidance to clinicians for selecting the most pertinent interventions relevant to a patient’s ARFID presentation. Much like the evidence-based out-patient care pathway, experts in the Radcliffe ARFID workgroup³ also agreed that not all patients with ARFID require a multidisciplinary treatment team. Rather, in addition to a mental health clinician, patients generally require a minimum of one other provider (e.g., pediatrician, primary care practitioner) for medical monitoring. For example, a patient with nutrition deficiencies would likely benefit from the expert counsel of a dietitian, and a patient with extremely low weight and inability to take in food orally would benefit consultation with a gastroenterologist to determine the appropriateness of placing a feeding tube. The Radcliffe approach is consistent with the ARFID treatment studies we describe here, where all except one study conducted at an inpatient program⁵⁵ relied on a mental health clinician with concurrent medical monitoring.

Overall, extant treatments for ARFID appear to produce favorable outcomes; however, they utilize small sample sizes (ranging from 11–60) comprising case reports and small case series. Importantly, only one study⁵³ included a comparison group, and the treatment as usual intervention was fairly minimal (primarily visits with a pediatrician). Larger, fully powered, randomized controlled trials with longer follow-up periods are critically needed to examine outcomes over time. Future research should also include dismantling studies to explore the relative efficacy of various treatment components (e.g., establishing a regular pattern of eating, systematic desensitization, interoceptive exposure) as well as treatment comparison studies. Some treatments are resource-intensive, requiring large teams and inpatient hospitalizations, whereas others occur over a brief, time-limited period of one session per week. Therefore, the cost-effectiveness of various treatments should also be examined. One potential avenue for future research is to develop digital mental health interventions, perhaps including a mobile app. Indeed, both CBT-AR and FBT-ARFID have been delivered via videoconferencing technology⁵⁷, engendering the possibility that treatments may be adapted to fit a digital modality.

Conclusions

Ten years following the induction of ARFID to DSM-5, several promising assessments and treatments are in the early stages of research. Indeed, structured clinical interviews and self-report measures are useful for assessing the presence and severity of ARFID, each with its own strengths and limitations that we outline in Table 1. Several novel treatments have been evaluated, albeit with modest sample sizes and mostly without comparison groups, necessitating future research using larger samples and randomized control trials. The most promising treatments with the highest level of evidence are CBT-AR, FBT-ARFID, and intensive multidisciplinary intervention. Despite advances in assessment and treatment, controversies remain. First off, the criteria for pediatric feeding disorder¹² overlap substantially with the criteria for ARFID. It is unclear whether it would benefit patients to have more than one diagnosis, or whether this creates unnecessary ambiguity. Experts in the feeding disorders and eating disorders have yet to arrive at clinical consensus. Strengths of two diagnoses include that some individuals may be captured under pediatric feeding disorder that would not be captured under ARFID. However, future research is needed to determine the necessity of having two sets of criteria to describe similar populations, especially given that the two disorders may respond to similar treatments. Second, the optimal method for ascertaining Criterion A2 (i.e., nutrition deficiencies) is unclear. Methods abound including self-report dietary recall, blood tests, and identifying signs/symptoms resulting from deficiencies. These varied methods could lead to patients being diagnosed differently depending on context, in turn potentially resulting in inconsistencies regarding eligibility for treatment. Third, because ARFID impacts so many different body systems and could theoretically be treated by many different disciplines, it is unclear which team members are required. This has implications for clinical care because different disciplines offer different types of treatment for ARFID (e.g., where a psychologist might offer exposure to new foods, occupational therapists might attempt to ameliorate feeding skills deficits). Depending on the presentation, some patients may be treated using a mental health clinician and concurrent medical monitoring only, whereas more severe patients may require a multidisciplinary team. Finally, though ARFID cannot be concurrently diagnosed with AN or BN, and DSM-5 precludes shape/weight disturbance for diagnosis, investigators^5,15,16 have documented overlap between ARFID and other eating disorder symptoms, and ARFID can co-occur with BED. Given the prevalence of normative body discontent^13,14, we encourage clinicians assessing and treating ARFID to exercise caution and utilize clinical judgment when conferring diagnosis. In sum, there is growing literature pertaining to the assessment and treatment of ARFID. Despite myriad advances in these areas since ARFID was introduced to the psychiatric nomenclature in 2013, future research is needed to improve understanding of the disorder and randomized control trials comprised of larger samples are warranted to examine treatment efficacy.