Avoidant/Restrictive Food Intake Disorder (ARFID): Two-Year Outcomes and Predictors of Persistence

Avoidant/Restrictive Food Intake Disorder (ARFID) is a newly recognized eating disorder, characterized by a persistent avoidance or restriction of food intake. Unlike anorexia nervosa, ARFID is not driven by body image or weight concerns. Instead, food restriction may result from heightened sensory sensitivity, a fear of aversive consequences, or a general lack of interest in eating. ARFID can lead to significant nutritional deficiencies, medical complications, and psychosocial impairment, and is distinct from “picky eating,” which is typically developmentally limited.

A recent prospective study from Evelyna Kambanis, PhD and colleagues Kendra Becker, PhD, Lauren Breithaupt, PhD, Kamryn Eddy, PhD and Jennifer Thomas, PhD from the Eating Disorders Clinical and Research Program at Mass General followed 100 youth and young adults (ages 9–23, 49% female; 91% White) with full or subthreshold ARFID over a two-year period. Participants were assessed at baseline, one year, and two years using the Pica, ARFID, and Rumination Disorder Interview (PARDI) to confirm subthreshold and full diagnoses at baseline and to evaluate symptom severity across ARFID presentations.

In this prospective longitudinal study, the investigators examined the persistence and remission of ARFID, patterns of diagnostic crossover, and predictors of outcome. 

Persistence and Remission of ARFID

Nearly half of the  participants retained their ARFID diagnosis across each follow-up point (45% at year one and 46% at year two), with 44% meeting criteria consistently across both years. Rates of remission were modest, occurring in 26% of participants at year one and 24% at year two.

Diagnostic crossover was rare. Only 3% of participants shifted to a restricting subtype of anorexia nervosa, suggesting that transition from ARFID to other eating disorders is uncommon compared to crossover patterns typically seen with other eating disorders.

ARFID Profiles and Predictors of Outcome

Previous studies have emphasized that ARFID is a heterogeneous disorder and have identified several recognizable profiles defined by unique mechanisms of food avoidance:

• Sensory Sensitivity: Restriction based on an aversion to certain tastes, textures, or smells.
• Fear of Aversive Consequences: Restriction due to fear of choking, vomiting, or food poisoning.
• Lack of Interest in Food/Eating: Consistent low appetite or disinterest in food.

In the current study, Kambanis and colleagues observed that symptom profiles predicted different outcomes:

• Greater sensory sensitivity and lack of interest at baseline were associated with worse outcomes, significantly increasing the likelihood of symptom persistence across the two years.
• In contrast, greater severity of fear of aversive consequences was linked to greater likelihood of remission at year two.

These findings suggest that profiles rooted in early-onset patterns (such as sensory aversions or low appetite) may contribute to enduring illness, while fears tied to negative experiences may be more modifiable.

Clinical Implications

This prospective study provides valuable insight into the natural course of ARFID in youth and young adults, demonstrating that the disorder often persists despite developmental changes and passage of time. Nearly half of individuals continued to meet criteria across two years, and only a minority experienced remission—underscoring that ARFID is not a transient phase of eating difficulty but rather an enduring condition requiring clinical attention. Importantly, persistence was observed not only in full-threshold ARFID but also in subthreshold presentations, suggesting that even mild cases may follow a prolonged trajectory if left untreated.

The study also highlights the importance of ARFID profiles in predicting outcomes:

• Individuals with sensory sensitivity and low interest in food at baseline were significantly more likely to experience persistence of ARFID symptoms. These features often develop early in life, are deeply entrenched, and may represent more enduring mechanisms that reinforce long-standing avoidance patterns.
• In contrast, greater severity of the fear of aversive consequences profile predicted higher likelihood of remission at two years. Since these symptoms are often precipitated by acute experiences (such as choking or vomiting), they may be more responsive to interventions targeting cognitions, behaviors, and anxiety related to eating.

Diagnostic crossover was rare overall, with only 3% of participants developing anorexia nervosa (restricting subtype). Importantly, those who shifted to anorexia tended to already be underweight and continued to endorse ARFID symptoms, suggesting significant diagnostic overlap and highlighting the complexity of differentiating between ARFID and anorexia in younger, low-weight populations.

Taken together, these findings reinforce that ARFID is clinically distinct from both picky eating and other eating disorders. Its persistence over two years—particularly among those with sensory sensitivities and low appetite—underscores the need for early recognition and specialized treatment tailored to individual symptom profiles. Clinicians can be reassured that diagnostic crossover to anorexia appears uncommon, yet should remain attentive to weight status and emerging body image concerns, as these may shift the clinical picture.

By recognizing the heterogeneity of ARFID and addressing specific drivers of food restriction, clinicians can improve treatment precision and potentially alter the long-term trajectory of the disorder. Future work should continue to disentangle how different ARFID subtypes evolve, and how interventions can be optimized to support recovery and prevent chronicity.

Other collaborators include Nassim Tabri, Iman McPherson, Julia E Gydus, Megan Kuhnle, Casey M Stern, Elisa Asanza, Kendra Becker, Lauren Breithaupt, Melissa Freizinger, Lydia A Shrier, Elana M Bern, Kamryn  Eddy, Madhusmita Misra, Nadia Micali, Elizabeth A Lawson, and Jennifer J Thomas.