This study adds to the literature on ARFID by comparing a cohort of children and adolescents undergoing day treatment for EDs, including patients with this “new” diagnosis. Notably, almost a quarter of our patients were diagnosed with ARFID, which illustrates the significant prevalence of this disorder amongst children and adolescents requiring an intensive level of ED treatment in a tertiary care setting. This was a higher prevalence than that found in the multicenter studies [14,15], which might be accounted for by the fact that our patients were encountered over four years in a day treatment setting, as opposed to all ED patients presenting for initial evaluation over a one-year period. The prevalence rate was in even starker contrast to the 11-year retrospective review from Canada, where the prevalence was only found to be 5% . There is no mention of age range in that study, only that the patients were adolescent ED patients assessed in a pediatric tertiary care hospital program. Our study included children and adolescents between 7 and 17 years, which may have been a slightly lower range than the Canadian study; this might also justify the higher prevalence of ARFID found in our cohort. Another possible explanation for the discrepancy in prevalence rates across studies is that younger patients with atypical EDs, like ARFID, may be increasingly referred to adolescent medicine ED programs in more recent years, as there has been greater recognition of these presentations as true EDs. The Canadian study reviewed records starting in 2000 and it would be interesting to know whether the prevalence increased annually over the 11 years. In our experience, referrals from primary care providers tend to generate more referrals once they are successfully managed. Lastly, the higher prevalence in our cohort may reflect the fact that many children and adolescents with ARFID present acutely and significantly malnourished, requiring a higher level of care, such as day treatment.
Similar to the multicenter and Canadian studies [15,16], our results demonstrate that there are significant demographic and diagnostic characteristics that differentiate children with ARFID from those with other EDs. First, while female patients remain the majority, there was a higher preponderance of male patients in the ARFID group than in the other ED groups. Children and adolescents with ARFID were more likely to present at a younger age with significant weight loss or failure to gain appropriate weight, were more dependent on oral or enteral nutritional supplementation, and had significantly more fears of choking and/or vomiting, and texture and/or sensitivity issues regarding food. These findings are consistent with those in studies of early-onset EDs [2,22,23], as well as in the recent multicenter study , and many are relevant and important features in making the diagnosis of ARFID .
Based on DSM-5 criteria, a patient cannot have body image distortion and be diagnosed with ARFID. However, our data revealed that 21% of patients diagnosed with ARFID had body preoccupation with somatic concerns. It is important to reiterate that none of the patients with ARFID had been diagnosed with AN using DSM-IV criteria, which underscores the absence of true body image distortion. During evaluation of a young patient with possible ARFID versus AN, it is critical to probe about body concerns that need to be distinguished from body image distortion. For example, if a patient has worries about becoming fat, this may have something to do with events in the family’s medical history, e.g. an overweight parent or grandparent with a recent myocardial infarction or diabetes diagnosis. Children and adolescents are often privy to this information, but may make illogical associations based on their cognitive developmental stage. This knowledge may then trigger restrictive eating behaviors. Thorough history-taking can often elicit this information.
As has been documented in other studies of patients with acute food avoidance without weight/shape concerns [2,15,22,25,26], there were no significant differences in our study between % MBW in patients with ARFID and AN; however, patients with AN lost a significantly greater percentage of their premorbid body weight. This may be explained by the fact that our patients with ARFID, notably those with the acute food refusal seen in functional dysphagia, may have presented sooner after the onset of illness than those with AN. The data may not fully bear this out due to the heterogeneity of the ARFID category (e.g. more chronic selective eaters vs more acute food refusal), which might balance out the length of illness data. Furthermore, young patients may present relatively early in the course of their illness, based on their age alone.
Based on both clinician and parental report, patients with ARFID had significantly more anxiety and less depression than patients with other EDs, which is similar to findings in the large multicenter study on ARFID . However, our study is the first of patients with ARFID to use standardized measures obtained from parents to aid in evaluation. There were no self-reported significant differences found between children with ARFID and those with other EDs on the RCMAS or any of its subscales, which could be due to the generally high comorbidity of anxiety symptoms in EDs. Alternatively, younger patients (those more likely to be diagnosed with ARFID) may have had a harder time filling out the questionnaire than older subjects, perhaps in understanding the questions or acknowledging symptoms of anxiety, due to cognitive developmental stage. It is important to clarify that ARFID is not simply a type of anxiety disorder, as the severity of the eating disturbance exceeds that which might be seen in an anxiety disorder and necessitates further clinical attention (see Table 1) .
Other than the use of outpatient psychotherapy, there were no significant differences between the groups in terms of prior mental health treatment, including hospitalizations for EDs or other mental health issues, admissions to day treatment programs, intensive outpatient programs, or residential treatment facilities. It should be taken into consideration, however, that ours is a young, relatively treatment-naïve population, and that the rate of past mental health admissions would be very different when looking at an older population of patients. Additionally, children with ARFID may be more likely seen as medically ill initially, and the early referrals may tend to gravitate toward the medical as opposed to mental health arena, as a trend in our data revealed, although it was not significant.
There were several strengths to this study, including the large sample size and the use of both clinical and standardized psychometric measures for patient assessment. Additionally, the use of multiple informants (patients, parents, and clinicians) adds to the validity of the findings. Furthermore, experienced clinicians completed all assessments and the adolescent medicine physician involved in deciding on the retrospective DSM-5 diagnoses was integrally involved in the efforts leading up to the inclusion of ARFID in the DSM-5. As ARFID is still a relatively “new” diagnosis, there are no formalized assessment tools available yet. However, instruments will likely be developed, capturing the clinical features and diagnostic criteria which will help standardize diagnosis. There are some available resources to help guide the clinician in evaluation [24,27].
However, there are several limitations that deserve mention. The retrospective nature of this study, and the fact that diagnoses were made on DSM-5 criteria that had not yet been formalized by the time of its completion, need to be taken into consideration. However, as previously mentioned, the published DSM-5 criteria were essentially the same as the proposed criteria used for this study. Careful discussion amongst experienced clinicians very familiar with all of the cases was undertaken to decide upon the appropriate DSM-5 diagnosis for each patient; this did not allow for direct assessment of inter-rater reliability. The absence of blinding of the clinicians may have introduced bias to the outcome of the study, possibly leading to a higher prevalence of ARFID than previously seen in other studies. Lastly, our patients were undergoing day treatment, which implies a certain severity of illness, and may limit the generalizability to patients in other settings, or non-clinical populations. Despite these limitations, this study provides support for ARFID as a separate diagnostic category.