Avoidant-restrictive food intake disorder (ARFID) was classified as a novel eating disorder (ED) in the DSM-5. Prior to that, any severe notion of picky eating in children was categorized as EDNOS, a DSM-IV notation that referred to any ED that was not anorexia nervosa or bulimia nervosa (Fisher et al., 2011). What patients meeting this classification had in common was no significant fear of weight gain or poor sense of body image, but, instead displayed a lack of interest in eating, expressed a sensory resistance, or feared physical consequences of eating, like choking or vomiting (Pinhas et al, 2016). In some instances, patients may not be underweight but likely lack in nutritional health (Nicely et al. 2014). Including ARFID as its own DSM-5 classification “improved clinical utility and captured a population of young people who had an eating disorder” (Fisher et al, 2011, p. 50) that previously was too undefined to warrant attention.
The population most impacted by ARFID is quite a vulnerable one: young children and early adolescence (Ornstein et al., 2012). Nicely et al. (2014) report picky eating in general to be a “common problem of childhood with anywhere from 13-22% of children between the ages of three and eleven...While young children are typically thought to ‘grow out of’ their pickiness, studies have shown that between 18-40% of the rigidity concerning food persists into adolescence” (p. 2). Also unique to note is the higher prevalence of males who meet ARFID criteria compared to other EDs (Fisher et al., 2011). While strict diagnostic criteria are still under investigation, broad strokes qualifications included the previously mentioned repulsion or disinterest in certain foods -- or only tolerating certain foods while strongly disregarding other foods -- or a fear of vomiting or choking that results in nutritional deficiency which is sometimes evaluated by substantial weight loss (Nicely et al., 2014). ARFID would not be a suitable diagnosis for a patient who suffers from nutritional deficiency or weight loss due to food scarcity or other medical issues (Nicely et al., 2014).
Given the relative newness of ARFID as a diagnostic possibility, studies related to intervention are relatively sparse, especially compared to other more widely researched EDs, like anorexia or bulimia. Even so, there is consistent success in the utilization of Family Based Treatment/Therapy (FBT). Lock et al. (2018) report that FBT’s adaptation to ARFID treatment 1) allows a de-emphasis on what caused the disorder (since that is largely unknown); 2) promotes a separation of the illness from both the family and the patient which alleviates notions of guilt or blame; 3) draws attention to the health consequences related to ARFID; 4) empowers parents to facilitate and participant in the change with and for their child; and 5) focuses on the dysfunctional eating behavior only, not the overall social dynamic between the patient and their family. While many other tools may be utilized in tandem with or under the umbrella of FBT -- especially in regard to behavior modeling, rewards systems, and in some instances prescription medication -- the general treatment plan is conducted in phases, one with a specialist (therapist) working closely with the whole family, offering special guidance for the parents to establish goals and feeding procedures (which may include implementing a rewards system or other agreed-upon strategies), and a second phase where the patient gains some autonomy in the decision-making process (Lock et al., 2018). The goal of this treatment is not to determine an internal family dysfunction as the cause of ARFID, but, rather, to coach the entire family on ways to improve the patient’s nutritional intake (Lock et al., 2018).
Assessment of Eight Related Studies
Study: Murray et al. (2012)
As with most of the studies related to ARFID treatment, Murray et al. (2012) focused on a single case as the basis of the research. The subject was Diego, a nine-year-old boy whose parents reported that he had displayed restrictive eating behaviors since he was two-years-old. At the start of treatment, his accepted food list contained three items and his body mass index (BMI) of 15.4 put him in the 32nd percentile for his age group (Murray et al. 2012). FBT treatment was initiated but deemed ineffective relatively quickly due to signs of distress from Diego coupled with no improvement in his nutritional intake or weight gain goals. A more general family therapy model was utilized in which Diego gave his food anxiety a name -- Beaster -- which allowed for deeper discussions related to Beaster’s role in Diego’s life. As treatment continued, Beaster became “smaller and less scary looking” (Murray et al., 2012, p. 273) while Diego simultaneously gained confidence in selecting foods on his own to try. At the end of his treatment -- a total of nine total sessions -- Diego recorded a BMI of 16.7, which moved him into the 58th percentile, a marked success (Murray et al., 2012).
In this study, non-probability sampling was used. As previously mentioned, this is a relatively new area of research, so many of the studies use very small sample sizes, if not a single case, as seen in Murray et al. (2012). Digging deeply into a case study such as Diego’s may not reflect the most robust evidence of external validity, but as other studies will show, the results in Murray et al. (2012) are informative of are informative of future studies.
For Diego’s case, the dependent variable (DV) is BMI and the independent variable is the non-resistant inclusion of novel foods into his diet: the more foods he added to his diet, the higher his BMI. The related interval measure is the percentile his rising BMI categorized him as in comparison to peers. Such percentiles are widely used for a number of different childhood milestones and growth rates, making them both reliable and valid. This reliability/validity gives them strength, though the weakness is that gaining weight does not necessarily correlate to higher nutritional intake. Other study limitations include an inability to attribute definitive success to the family-based therapy and lack of knowledge about if Diego will continue to increase his novel food intake on his own or not.
Conceptually, using FBT to treat ARFID is a relatively new intervention that could lead to deeper understanding of which evidence-based practice(s) best serves patients with this diagnosis. Operationally, this study had simple terms for deciding which variables should be manipulated and which would indicate the success, failure or neutrality of the intervention. Considering the study design, while it is true that the FBT itself needed to be dialed back a bit for Diego’s case, even the “relaxed” version of treatment still proved quite effective in yielding a desirable outcome. The success is evident in such an outcome and also in the adaptability of both the therapist and the family to alter their course of action slightly to accommodate Diego’s personalized needs. As such, the results have only marginal external validity since the researchers did not hold steady to their planned treatment and, instead, customized the process for the patient.
Study: Sira and Fryling (2012)
Like the previous study, Sira and Fryling (2012) utilized a single case study to examine a avoidant/restrictive food intake case that is very similar to Murray et al. (2012). Here, researchers studied Desmond, age nine, who displayed nearly identical behaviors as Diego. This case did not report Desmond’s BMI, perhaps making him a case of an ARFID patient who did not display severe BMI deficiency, though that is not explicitly stated. Instead, in this case, the therapist and the family decided to try Peer Mirroring where Desmond sat across a table from his six-year-old sister (a “normal” eater) to have her model desirable eating behaviors. For this experiment, eating sessions were timed and success was determined by whether or not Desmond took a bite of his food and swallowed it within thirty seconds. Desmond’s parents worked out a rewards system for each new food he tried. In the second phase, Desmond went through this ritual without his sister’s participation. The result was a 100% improvement rate on the foods his parents had selected for him to incorporate into his diet (Sira and Fryling, 2012).
Many of the assessment points are identical to Murray et al. (2012): nonprobability sampling was used, which makes the sample strong for the singular case of Desmond but more diluted terms of immediate broader application. The IV in this study is once again the novel foods being introduced but the DV is related to “mouth clean” or Desmond’s ability to swallow the novel food in the allotted time. This nominal measurement is clearly defined as his mouth being empty of food, an observable fact, making it both reliable and valid. Conceptually, the collaborative nature of the treatment paved the way for additional studies to be carried out. Operationally, this study had simple terms for determining a productive or nonproductive outcome. The study design in this case needed no adjustment and was carried through with striking results. While the external validity of those results is less viable across the spectrum of ARFID cases, Desmond’s results indicate a positive correlation between a family-direct treatment and a desirable outcome. Study limitations are the same as the previous case, except that in this instance, no alterations were needed for the study design.
Study: Murphy and Zlomke (2016)
This study is once again a single case. The patient is a six-year-old named Molly with a history of gastroesophageal reflux disease (GERD) that caused frequent vomiting for the first year of her life. As a result, Molly began refusing previously accepted foods as a nine-month-old and developed other abnormal eating behaviors, like excessive chewing. By the time she was in kindergarten, her BMI ranked in the 81st percentile. At the time of her treatment, Molly had 10-15 reported foods that she would willingly consume (Murphy and Zlomke, 2016).
The therapist in this instance had Molly’s mother complete the BPFAS, a widely used measure of pediatric eating behaviors, and a fear hierarchy chart to assess her baseline and chart progress. Molly’s mother was coached to offer positive reinforcement only during regulated feeding sessions and a rewards system was implemented. At the end of 18 sessions held over a six month period, Molly’s BPFAS score decreased from 89 (considered clinically elevated) to 71 (considered a normal limit), her feeding related problems decreased from a score of 13 to 2 , and she was accepting an average of 20-30 new foods per week (Murphy and Zlomke, 2016).
Like the previous two studies, nonprobability sampling was used, making Molly’s results anecdotal to broader usage despite their undeniable success for her specific case. The IV in this study is once again the novel foods being introduced but the DV is the BPFAS score. This interval measurement is a clearly defined diagnostic tool. Conceptually, the collaborative nature of the treatment paved the way for additional studies to be carried out. Operationally, this study had simple terms for determining a productive or nonproductive outcome. The study design used a potentially more reliable DV in that BPFAS is commonly used for food intake assessment in pediatric cases. While the external validity of those results is less viable across the spectrum of ARFID cases, Molly’s results indicate a positive correlation between a family-based treatment and a desirable outcome. Study limitations are the same as the previous case.
Study: Ornstein et al. (2017)
This study was a retrospective review of 177 participants in a partial-hospitalization program (PHP) between 2008 and 2012 that worked with ARFID patients between the ages of 7 and 17 (Ornstein et al., 2017). Most of the patients expressed fear-based reasons to restrict their eating (such as fear of choking) or a general disinterest in food, while a few fit the third ARFID classification of sensory reasons to avoid eating. Of the 177 charts originally under consideration, 130 remained viable for the duration of the study. These patients and their families worked closely with therapists to establish healthier eating behaviors by regulating meal time procedures and implementing rewards systems. In some instances, psychotropic medications were also utilized.
The metrics for analysis included the patients’ BMI as well as assessment by both the Children’s Eating Attitude Test (chEAT) and the Revised Children’s Manifest Anxiety Scale (RCMAS). Analysis of the charts revealed that ARFID patients had significantly shorter lengths of stay in the program compared to anorexia patients and, overall, ARFID patients showed the greatest improvement compared with program participants diagnosed with alternative EDs. This study also cited the flexible use of FBT alongside other interventions, including psychotropic medications, as a strong reason for the success noted in ARFID cases.
Again, non-probability sampling was used. In this case, ARFID patients were compared to patients with other EDs undergoing the same treatment and the results showed strong rates of success for ARFID patients, which boosts the overall reliability and validity of this course of treatment for the considered population. The IV in this study was the ED diagnosis and the DVs were BMI, chEAT scores, and RCMAS scores. These interval measurements are clearly defined diagnostic tools. Conceptually, this broader testing sample creates higher confidence in the overall success rate for treatment when applied to the general categorization of ARFID patients. Operationally, it revealed the faster favorable response of ARFID patients in comparison to those with other EDs. The study design is more comparative of this intervention approach being used across three different ED diagnoses -- ARFID, anorexia, or bulimia -- than previously discussed studies, and, as such, reveals how highly treatable ARFID is, creating a stronger external validity for FBT-associated interventions with regard to these patients. Study limitations include no follow up, no control group, and no definitive determination for which aspects of the program were the most successful.
Study: Lock et al. (2018)
This study looked at three ARFID cases: Lilliana, age 8; Allison, age 9; and Isabella, age 11. For each patient, baseline assessment was recorded via reported scores from Pica, ARFID, Rumination Disorder Interview (PARDI) and their weight prior to intervention. In all three cases, therapists worked with the patients and their families to discuss the importance of caloric intake, planned constructive meal time rituals, and moved on to the slow incorporation of novel foods into the patients’ diets. By the end of treatment, all three patients showed improved PARDI scores and appropriate weight gains (Lock et al., 2018).
Again, a non-probability sample was used and the results proved highly successful for the participants and anecdotal to the broader field of ARFID patients. The IV is the novel foods being introduced and the DVs are PARDI scores and weight. These interval measurements are clearly defined diagnostic tools. Conceptually, the collaborative nature of the treatment paved the way for additional studies to be carried out. Operationally, this study had simple terms for determining a productive or nonproductive outcome. The study design used a diagnostic measure to assess results, but even so, the external validity of those results remains largely speculative with such a small sample size. This study emphasized the limitation of a lack of clear, clinical diagnostic criteria for ARFID.
Study: Spettigue et al. (2018)
This study focused on six ARFID cases (five female, one male) with a median age of 12.9 and an average assessment of 80.5% of their goal weights (Spettigue et al, 2018). Treatment included FBT alongside other interventions, including psychotropic medications. Therapists worked with the patients and the families to establish structured meal times aimed to reduce food-based fear and anxiety. Parents were coached to offer positive reinforcement for desired behaviors and set up a rewards system. By the end of treatment, all six patients had achieved their goal weight.
Once more, nonprobability sampling was used, making the results anecdotal to broader application. The IV is novel foods and the DV is weight, an interval measurement. Conceptually, this study further strengthens the results from similar studies and operationally, its terms are straightforward in their analysis. The study design shows the positive impact of FBT-driven intervention on ARFID patients but is inconclusively the reason for the desired results since medications were also used. The external validity of the study is more impactful when combined with the results from other studies but on its own, it stands out most as encouragement for further studies to be conducted.
Study: Taylor et al. (2018)
Returning to the single case study format, Taylor et al. (2018) reports the findings of family-based treatment for an ARFID patient named Haydell, age 13. Prior to intervention, Haydell voluntarily consumed seven different foods (no fruits or vegetables) and was self-identifying concern of nutritional deficiency. Alongside her parents and her therapist, target novel foods were identified and slowly integrated into the family’s meals. Haydell’s progress was measured by her ability to consume the novel foods in a minimum of a pea-sized quantity and then display “mouth clean” by allowing the inside of her mouth to be visually inspected as empty. By the time of session termination, Haydell had reached a 100% success rate in her novel food consumption and had increased her voluntary food selection from 7 foods to 61 (including raw vegetables). The researchers followed up via email nine months after treatment and saw no signs of backsliding (Taylor et al., 2018).
This study used non-probability sampling. Haydell’s results are a tribute to her cited personal desire to improve her nutrition but have anecdotal implications of a successful intervention for ARFID patients. The IV is once again the novel foods being introduced and the DV is an achievement of “mouth clean.” This nominal measurement is practical. Conceptually, this treatment suggests a successful path for other ARFID cases. Operationally, this study had simple terms for determining a productive or nonproductive outcome. The study design used a changing criterion platform that started with Haydell consuming three bites per session before ramping up to five bites per novel food. This mutually agreed-upon increased exposure allowed for Haydell and her parents to build her positive meal time behaviors gradually. The external validity of Haydell’s case may be too singular to signify an immediate success for ARFID patients, but it certainly warrants further investigation. The study limitation is a lack of knowing if Haydell’s maturity played a significant role in her recovery or if the intervention itself was the reason for her positive result.
Study: Rienecke et al. (2020)
This study focused on three ARFID cases participating in a PHP treatment reliant on FBT principles. The patients, an eight-year-old female, a ten-year-old male, and a fourteen-year-old male, all worked in tandem with their parents and their therapists to develop healthy eating strategies aimed to increase the number of novel foods they would willingly consume. The parents were coached in relaxation and encouragement strategies geared towards easing their child into a positive outcome. Each patient filled out the Children’s Depression Inventory (CDI) and Multidimensional Anxiety Scale for Children (MASC) at intake. By the end of treatment, the eight-year-old female had gained 4.9 lbs, her CDI score went from a 4 to a 5, and her MASC score went from a 45 to a 24; the ten-year-old male gained 3.6 lbs, though he did not fill out the CDI or MASC during his termination so there is no comparison available; and the fourteen-year-old male gained 18.2 lbs, saw his CDI score shift from a 3 to a 1, and his MASC score drop from 19 to 18 (Rienecke et al., 2020).
As with the other studies discussed, non-probability sampling was used, making the results anecdotal to broader application. The IV is novel foods and the most consistent DV across all three cases is weight, an interval measurement. Conceptually, this study seemed less convincing than some of the other studies in that the metrics of measurement discussed seemed inconsistently applied or explained. While the goal of educating the parents to take charge in their child’s feeding behavior appears to be met, the patients themselves seemed to make much more incremental progress than some of the other studies this analysis has explored. The external validity of the study is mostly anecdotal. A limitation in this study is a lack of ARFID-specific assessment metrics that cause researchers to rely on weight -- which is not necessarily an indicator of nutrition -- to assess intervention success or failure.
Application to Disadvantaged Populations
Due to the relatively new nature of ARFID-related research and the almost universal reliance on single or minimal (under ten) research participants per study, there was no discussion about implications for ARFID patients across social platforms related to race, ethnicity, socio-economic status, or anything related. While the genders and ages were uniformly reported, either directly or as an average, other social factors were generally not featured in case analysis and no studies discussed noticeable differences related to those factors. More research is needed to determine if there is a causal or significant relationship between populations considered to be advantaged versus disadvantaged in relation to an ARFID diagnosis.
Little is truly known about the causal onset of EDs in children and with ARFID in particular, diagnostic criteria is still quite broad (Pinhas et al, 2016). As such, clinicians are navigating treatment for their patients by looking to FBT, an intervention with notable success for other common EDs (Spettigue et al., 2018). Taylor et al. (2018) report that “the use of [FBT] strategies in feeding interventions includes participants talking about their fears, anxieties, and beliefs around nonpreferred foods or participating in...family therapy sessions” (p. 12). With no clearly established “go to” treatment for ARFID, FBT is increasingly utilized as the research shows consistently that parent education about how to establish consistent, healthy eating rituals is quite significant (Rienecke et al., 2020). Rienecke et al. (2020) write, “Parents...benefit from the expertise of a specialist who can analyze the unique driving factors that perpetuate an unhelpful eating behavior in children with ARFID” (p. 301).
The studies reported on here are certainly a good start but there is a long way to go to understand not only ARFID as a diagnosis but discovering broadly applicable interventions that will reduce or eliminate the patterns of this ED in the young population it generally afflicts. Establishing clear diagnostic criteria as well as a more consistent basis of evaluating the success or failure of the applied intervention are still needed. That said, the overall reports on the utilization of FBT in ARFID patients offers a great deal of hope, since nearly all of the cases saw positive results from this intervention.
Fisher, M. M., Rosen, D. S., Ornstein, R. M., Mammel, K. A., Katzman, D. K., Rome, E. S., ... & Walsh, B. T. (2014). Characteristics of avoidant/restrictive food intake disorder in children and adolescents: a “new disorder” in DSM-5. Journal of Adolescent Health, 55(1), 49-52.
Lock, J., Robinson, A., Sadeh‐Sharvit, S., Rosania, K., Osipov, L., Kirz, N., ... & Utzinger, L. (2019). Applying family‐based treatment (FBT) to three clinical presentations of avoidant/restrictive food intake disorder: Similarities and differences from FBT for anorexia nervosa. International Journal of Eating Disorders, 52(4), 439-446.
Murphy, J., & Zlomke, K. R. (2016). A behavioral parent-training intervention for a child with avoidant/restrictive food intake disorder. Clinical Practice in Pediatric Psychology, 4(1), 23.
Murray, S. B., Thornton, C., & Wallis, A. (2013). Selective eating in a 9-year-old boy: family therapy as a first-line treatment. Clinical child psychology and psychiatry, 18(2), 270-275.
Nicely, T. A., Lane-Loney, S., Masciulli, E., Hollenbeak, C. S., & Ornstein, R. M. (2014). Prevalence and characteristics of avoidant/restrictive food intake disorder in a cohort of young patients in day treatment for eating disorders. Journal of eating disorders, 2(1), 21.
Ornstein, R. M., Essayli, J. H., Nicely, T. A., Masciulli, E., & Lane‐Loney, S. (2017). Treatment of avoidant/restrictive food intake disorder in a cohort of young patients in a partial hospitalization program for eating disorders. International Journal of Eating Disorders, 50(9), 1067-1074.
Ornstein, R. M., Rosen, D. S., Mammel, K. A., Callahan, S. T., Forman, S., Jay, M. S., ... & Walsh, B. T. (2013). Distribution of eating disorders in children and adolescents using the proposed DSM-5 criteria for feeding and eating disorders. Journal of Adolescent Health, 53(2), 303-305.
Pinhas, L., Nicholls, D., Crosby, R. D., Morris, A., Lynn, R. M., & Madden, S. (2017). Classification of childhood onset eating disorders: A latent class analysis. International Journal of Eating Disorders, 50(6), 657-664.
Rienecke, R. D., Drayton, A., Richmond, R. L., & Mammel, K. A. (2020). Adapting treatment in an eating disorder program to meet the needs of patients with ARFID: Three case reports. Clinical Child Psychology and Psychiatry, 25(2), 293-303.
Sira, B. K., & Fryling, M. J. (2012). Using peer modeling and differential reinforcement in
the treatment of food selectivity. Education and Treatment of Children, 35(1), 91-100.
Spettigue, W., Norris, M. L., Santos, A., & Obeid, N. (2018). Treatment of children and adolescents with avoidant/restrictive food intake disorder: a case series examining the feasibility of family therapy and adjunctive treatments. Journal of eating disorders, 6(1), 20.
Taylor, T., Haberlin, A., & Haberlin, J. (2019). Treatment of avoidant/restrictive food intake disorder for a teenager with typical development within the home setting. Journal of adolescence, 77, 11-20.
Research paper written in pursuit of my MSSA (2020)
Virtual Tip Jar: Venmo @sarahwolfstar