INTRODUCTION
Avoidant/restrictive food intake disorder (ARFID) is a feeding and eating disorder (ED) included in the Diagnostic and Statistical Manual of Mental Disorders-Fifth Edition (DSM-5) [1]. ARFID is characterized by three distinct eating patterns; avoidance of foods due to sensory properties (e.g., picky eating), reduced appetite or lack of interest in eating, and fear of negative consequences from eating, such as choking or vomiting [1]. Although it replaced the DSM-IV category of feeding and EDs of infancy and early childhood [2], it can also occur in adolescents and adults [3].
Recent studies indicate that approximately 0.3% to 15.5% of children and adolescents [4] and between 0.3% and 2% of adults [5] can be diagnosed with ARFID. Variability in symptoms across age groups may be important in both adult and pediatric patients with ARFID. One study found that restrictive eating behaviors are less frequently observed in adults with ARFID compared to children and adolescents with ARFID [5]. Also, in a descriptive study, adults with ARFID often reported gastrointestinal sensitivity, a tendency to eat less when experiencing emotional distress, and comorbid anxiety and depressive disorders [6]. These findings highlight the importance of age-targeted treatment for pediatric and adult patients with ARFID.
ARFID and anorexia nervosa-restricting type (RAN) share restrictive eating behaviors, leading to potential nutritional deficiencies [7]. However, patients with ARFID lack the preoccupation with body weight/shape found in patients with RAN [1,8]. Clinical studies indicate an earlier onset of illness among patients with ARFID and a higher proportion of male patients compared to those with anorexia nervosa (AN) [9]. Additionally, studies have reported lower levels of ED and general psychopathology, including body image disturbances, depression, and anxiety, among adults with ARFID compared to those with AN [10].
In terms of body mass index (BMI), findings have been inconsistent. Some studies have reported lower BMIs in patients with ARFID [11,12], while others have reported higher BMIs [13]. Individuals with ARFID often present with lower premorbid and maximum BMIs, as well as higher minimum BMIs compared to those with AN [14]. This inconsistency highlights important distinction between ARFID and AN.
The comparison between ARFID and AN is further complicated by differences in psychiatric comorbidity. Anxiety disorders are highly comorbid with ARFID, with prevalence rates ranging from 21.4% to 50%, while comorbidity with depressive disorders ranges from 7.2% to 33% [4]. Children with AN exhibit higher levels of depression, drive for thinness, and overall ED psychopathology than children with ARFID [12]. Regarding psychiatric comorbidity, pediatric patients with ARFID have a higher incidence of anxiety disorders, whereas depressive disorders are more prevalent in those with AN [12]. These features distinguishing ARFID from AN in children are generally consistent with findings in adults [5,12]. However, specific differences in symptom severity and frequency between pediatric and adult patients with ARFID have been noted [6]. To our knowledge, no studies have directly compared pediatric and adult patients with ARFID.
Personality traits could also provide insight into the etiological and maintenance factors of EDs. Within the personality framework of the Five-Factor Model, individuals with RAN exhibit higher neuroticism than controls as well as higher conscientiousness and agreeableness compared to those with the binge-purge subtype of AN [15]. These personality traits help identify emotional responses and behavioral tendencies, allowing for the development of tailored treatment plans [15].
Meanwhile, behavior inhibition system and the behavior activation system (BIS/BAS) insights reveal how patients experience and react to symptoms, which is crucial for developing effective prevention and treatment strategies [16,17]. In the framework proposed by Gray’s behavior and personality system [18], under-reactivity and over-reactivity of the BIS and the BAS may be associated with eating behaviors. Increased BIS reactivity has been linked to unhealthy weight loss [16], as reduced calorie consumption is perceived as rewarding [17]. Additionally, a lack of BAS reactivity has been associated with RAN [17], further distinguishing RAN from other EDs.
Understanding personality traits and the BIS/BAS system is essential for developing personalized treatments. Personality traits help identify emotional responses and behavioral tendencies, enabling tailored treatment plans [15]. However, research examining personality traits in relation to ARFID remains limited [19].
Limited studies have been conducted on ARFID in non-Western cultures. The studies from non-Western countries suggest that ARFID might manifest with unique characteristics stemming from cultural contexts [14,20,21]. For example, in Japan, abdominal pain, rather than choking or vomiting, was reported to be the primary reason for food avoidance [14], whereas in China, the public and performative nature of meals might intensify the anxiety and social constraints associated with dietary restrictions in ARFID [21]. Studies conducted in Korea [22] and Singapore [23] indicate that picky eating and challenges with feeding might be more prevalent in certain non-Western contexts.
The current study aimed to explore the clinical features and personality traits of adult and pediatric patients with ARFID, as well as those with RAN in a Korean hospital cohort. Our hypotheses were as follows: 1) patients with ARFID would have an earlier age of onset and longer duration but less severe ED psychopathology compared to those with RAN; 2) ARFID would be associated with less severe comorbid psychopathology and neurotic traits compared to RAN; 3) adult patients with ARFID would have more severe comorbid psychopathology than pediatric patients with ARFID; and 4) there would be no differences in personality traits between adult and pediatric patients with ARFID.
METHODS
Participants
The retrospective chart review was conducted in a cohort of patients, aged <60 years, who sought treatment for an ED at a clinic in Seoul Paik Hospital in Seoul, South Korea between 2013 and 2023. The study involved 392 patients diagnosed with ARFID (n=68) and RAN (n=324). For the pediatric group (age <18 years old), 51 (75%) patients were in the ARFID group, and 190 (58.6%) patients were in the RAN group. This study was approved by the Institutional Review Board of Inje University (INJE 2023-12-019).
Measures
Eating Disorder Examination Questionnaire
The Eating Disorder Examination Questionnaire (EDE-Q) [26] is a 36-item self-reported questionnaire assessing the severity of ED psychopathology in the past 28 days on four subscales (restraint, eating concern, shape concern, and weight concern). The reliability of the Korean version is satisfactory (Cronbach’s α=0.92) [27].
Neuroticism-Extraversion-Openness Five-Factor Inventory
The Neuroticism-Extraversion-Openness Five-Factor Inventory (NEO-FFI) [30] is a 60-item self-reported questionnaire on five personality factors. The five domains measure neuroticism, extraversion, openness, agreeableness, and conscientiousness. The reliability of the Korean version is acceptable (Cronbach’s α=0.68-0.86) [31].
BIS/BAS
The BIS/BAS [32] Scale is a 24-item self-reported measure that assesses two motivational systems based on responsiveness to punishment and reward. The BAS scale is divided into three subscales (reward, drive, and fun-seeking). The Korean version has satisfactory reliability (Cronbach’s α=0.78-0.87) [33].
Procedure
Data were collected from the routine clinical assessments of patients during their initial visit to the clinic. They were diagnosed by an ED specialist based on DSM-5 diagnostic criteria, which was followed by a EDE interview with a clinical psychologist, and the results were then applied to the DSM-5 criteria. Patients completed self-reported questionnaires, including the EDDS [24] and EDE-Q [26] for assessing ED psychopathology, DASS-21 [28] for assessing general psychopathology, and NEO-FFI [30] and BIS/BAS scale [32] for assessing personality traits. They also completed a demographic questionnaire and provided information on ED history, including family history of ED, age at onset, highest and lowest BMI, the presence of self-harming behaviors, and the degree of weight suppression and weight suppression at lowest (kg). Weight suppression was measured as the difference between the highest-ever weight and current weight, while weight suppression at lowest was defined as the difference between the highest-ever and lowestever weight. Height and weight measurements were taken while patients wore light clothing and no shoes or personal belongings.
Data analyses
Differences in demographic and clinical characteristics between patients with ARFID and RAN, as well as between pediatric and adult patients in each disorder, were retrospectively investigated using independent samples t-test for continuous variables and the chi-squared test for categorical variables. The Mann-Whitney U test was conducted when the assumption of normal distribution was violated. Effect sizes were calculated as Cohen’s d for the t-test with continuous variables, Phicoefficients (Φ) for the chi-squared test with categorical variables, and rank-biserial correlation for the Mann-Whitney U test. The estimation of Cohen’s d is as follows: negligible (d=0 and <0.15), small (d≥0.15 and <0.40), medium (d≥0.40 and <0.75), large (d≥ 0.75 and <1.10), very large (d≥1.10 and <1.45), and huge (d≥1.45). Phi-coefficients are categorized as: small (φ≥0.1), medium (φ≥0.3), and large (φ≥0.5) [34]. Rank-biserial correlation effect sizes are interpreted as: small (rrb≥0 and <0.1), medium (rrb≥0.1 and <0.3), and large (rrb≥0.3) [35]. All analyses were conducted using SPSS 25.0 for Windows (IBM Corp.) with an alpha level of 0.05.
Comparisons of demographic and clinical characteristics between patients with RAN and ARFID
Table 1 presents the demographic and clinical characteristics of patients with ARFID and RAN. Both patients with RAN and ARFID were predominantly female, with ARFID patients having a higher male ratio (15.5%) compared to those with RAN (3.1%; χ2=18.828(1), p<0.001, φ=0.225). There were no significant differences in the mean age between the two groups. Additionally, there were no significant differences in the percentage of patients with a family history of EDs between the groups. Patients with ARFID had an earlier onset age compared to patients with RAN (t(353)=-3.602, p<0.001, d=0.54) and tended to have a longer duration of the disorder before presenting to the clinic (t(334)=2.013, p=0.045, d=0.31). There were no differences in the current BMI or the lowest-ever BMI between the groups, but the highest-ever BMI was lower in patients with ARFID (t(382)=-3.710, p<0.001, d=0.51).
In terms of ED psychopathology, patients with ARFID reported lower scores on both subscales and global scores compared to those with RAN, with large effect sizes (all p-values <0.001; Cohen’s d ranged from 1.24 to 1.51). Regarding general psychopathology, there were no differences in the percentage of patients reporting self-harm. However, patients with ARFID reported lower scores on the DASS-21 depression (t(234)=-3.513, p=0.001, d=0.65), anxiety (t(234)=-3.371, p=0.001, d=0.63), and stress subscales (t(234)=-4.446, p<0.001, d=0.82), and total scores (t(234)=-4.210, p<0.001, d=0.78) with medium to large effect sizes, compared to those with RAN.
In terms of personality traits, patients with ARFID reported lower scores on the BIS (t(151)=-2.216, p=0.028, d=0.51), but showed a similar pattern in the BAS scores, including reward responsiveness, drive, and fun-seeking, compared to those with RAN. On the NEO-FFI assessment, patients with ARFID exhibited lower levels of neuroticism (t(172)=-4.028, p<0.001, d=0.90) and higher levels of openness (t(172)=2.876, p=0.005, d=0.64) with large effect sizes.
Comparison of clinical and personality characteristics between patients with RAN under 18 and above 18 years old
Table 2 presents the comparison of the clinical characteristics between pediatric and adult patients with RAN. Regarding ED psychopathology, patients under 18 years old showed higher restraint scores on the EDE-Q compared to those above 18 years old (t(210)=2.525, p=0.012, d=0.35) with a small effect size, with no significant differences in the other subscales. For general psychopathology, patients under 18 years old with RAN reported lower depression (t(200)=-3.945, p<0.001, d=0.56), anxiety (t(200)=-2.283, p=0.024, d=0.33), stress (t(200)=-3.518, p=0.001, d=0.50), and total DASS-21 scores (t(200)=-3.712, p<0.001, d=0.53) with small to medium effect sizes, compared to adults. Regarding personality traits, no differences were observed in the BAS/BIS scores between the two groups. Similarly, there were no significant differences in NEO-FFI personality traits between the groups.
Comparison of clinical and personality characteristics between patients with ARFID under 18 and above 18 years old
Table 3 demonstrates a comparison of the clinical characteristics between pediatric and adult patients diagnosed with ARFID. The onset age was earlier (t(50)=-6.282, p<0.001, d=1.96), and the duration of the disorder was significantly shorter (t(46)=-4.139, p<0.001, d=1.31) in patients under 18 years old. Patients under 18 years old with ARFID had a lower current BMI (t(64)=-2.281, p=0.026, d=0.64) and highest-ever BMI (t(62)=-3.275, p=0.002, d=0.93) compared to adults with ARFID. Regarding ED psychopathology, there was no significant difference in the EDE-Q scores between the groups. For general psychopathology, patients under 18 years old with ARFID reported lower anxiety (t(32)=-2.602, p=0.014, d=0.89), stress (t(32)=-3.067, p=0.004, d=1.05), and DASS-21 total scores (t(32)=-2.819, p=0.008, d=0.97) with large effect sizes, compared to adults. In terms of personality traits, no significant differences were observed in the BAS/BIS and NEO-FFI scores between the groups.
DISCUSSION
The present study investigating ARFID and RAN in a Korean hospital cohort reported the clinical features of ARFID specifically in patients under 18 years old and adult patients. Overall, patients with ARFID had lower levels of ED psychopathology, depression, anxiety, and stress compared to those with RAN. They also reported lower levels of behavioral inhibition and neuroticism, and higher levels of openness compared to those with RAN. Patients under 18 years old with RAN reported higher levels of restrained eating compared to adult patients but had ssimilar personality profiles. Additionally, patients under 18 years old with RAN showed higher levels of restricted eating and lower levels of general psychopathology. Patients with ARFID under 18 years old showed less severe general psychopathology but had a personality trait profile similar to that of adult patients.
While the diagnostic differentiation between ARFID and RAN has traditionally focused on body image concerns, our study revealed multiple dimensions of variation. For example, patients with ARFID had an earlier onset and longer illness duration compared to those with RAN. This early onset and longer duration suggest a persistent pattern of restrictive eating in ARFID, consistent with previous research [36]. Additionally, patients with ARFID had lower highest-ever BMIs compared to those with RAN, which may be associated with consistent avoidance behaviors before the onset of the disorder [37]. Meanwhile, a higher proportion of males was found in patients with ARFID compared to those with RAN (15.5% vs. 3.1%), consistent with previous studies [9].
Patients with ARFID demonstrated lower levels of ED psychopathology and dietary restraint compared to those with RAN. This finding contrasts with the results of Nakai et al. [14], which involved a non-Western sample and reported no significant differences in the proportion of restraint behaviors between ARFID and AN. This result may stem from differences in measurement. While Nakai et al. [14] evaluated general restraint behaviors, the current study used the EDE-Q, specifically targeting “restraint behaviors aimed at weight loss.” Given that patients with RAN are more preoccupied with body weight/shape [1,8], EDE-Q may provide a more sensitive measure of weight loss targeted behaviors among them.
Furthermore, patients with ARFID in this study exhibited lower levels of anxiety, depression and stress compared to those with RAN. This finding indicates that patients with ARFID in our sample experienced less severe negative affect. Consistently, Nakai et al. [14] also reported that patients with ARFID displayed less severe negative affect than those with AN, supporting the notion that ARFID is associated with milder emotional distress relative to other EDs in non-Western sample.
In comparisons of personality traits between ARFID and RAN, patients with ARFID reported lower levels of neuroticism and higher levels openness than those with RAN. Individuals with lower levels of neuroticism tend to be less susceptible to negative emotions like anxiety, depression, and emotional instability. While neuroticism is commonly observed in patients with AN and has been reported to drive restrictive eating behaviors [15,37], ARFID may be less associated with it. Consistent with previous studies [3], patients with ARFID reported reduced behavioral inhibition compared to those with RAN, indicating that ARFID might be driven by an immediate reward of avoidance [7]. The traits of low neuroticism and reduced behavioral inhibition are in line with the relatively lower anxiety levels observed in patients with ARFID compared to those with AN. These traits may contribute to less emotional eating [38]. In addition, the results have clinical implications for the therapeutic potential of refeeding with less anxiety about food or weight among patients with ARFID. On the other hand, the results suggest that patients with ARFID may delay acknowledging the severity of their symptoms, allowing avoidant eating behaviors to persist over time [6,13,19]. Therefore, effective treatment for ARFID may require strategies such as external reinforcement and gradual exposure, along with individualized approaches depending on the patient’s motivation level and the underlying reasons for food avoidance [39]. There were no significant differences in personality traits between the pediatric and adult patients in the ARFID and RAN groups, which suggests that personality traits remain across development in both groups.
With regard to age differences in patients with RAN, eating pathology between pediatric and adult patients was not significantly different. Similarly, there were no significant differences in personality traits between pediatric and adult patients with RAN. Whereas pediatric patients with RAN exhibited less severe negative affect, including depression, anxiety, and stress, compared to adults with RAN, personality traits remained consistent across age groups. The mechanisms by which personality traits contribute to RAN, and how these traits can be effectively targeted in the treatment of AN need to be further investigated.
Concerning age differences in the ARFID group, pediatric patients reported less severe depression, anxiety, and stress than adults. This is consistent with the findings that ARFID in adults is associated with more severe psychopathology [19]. Our findings support a framework for developing age-specific interventions for ARFID. For pediatric patients with ARFID, the focus needs to be on correcting restrictive eating behaviors through interventions targeting key etiological factors, including sensory perceptions. Approaches like behavioral therapy focusing on the etiological model of sensory perception, appetite, and fear-based avoidance are known to be effective in treating pediatric patients. On the other hand, for adult patients with ARFID, interventions targeting psychological difficulties and chronic restraint eating seem crucial to enhance adaptability. It is important to consider the long-term effects of ARFID progression, as well as the severe psychological problems associated with the disorder [38].
With respect to cultural implications, our findings suggest a potential influence of cultural factors on ARFID presentation in Korea. Among patients with ED presenting to the clinic over a 10-year span, 3.2% were diagnosed with ARFID, which was significantly lower compared to RAN (15.2%). This finding differs from Western populations, where the lifetime prevalence of ARFID is comparable to that of AN [40]. The relatively low prevalence of ARFID in our cohort could be due to underrecognition of ARFID as an ED, which could potentially lead to fewer patients seeking treatment at the clinic.
The present study has strengths in examining a large Korean clinical cohort. However, there are limitations that should be considered when interpreting the results. First, we could not utilize a diagnostic tool or questionnaire for ARFID assessment: instead, we relied on DSM-5 criteria. Additionally, we were unable to include a healthy control group. Also, limitations regarding its retrospective cohort design (e.g., loss of data) need to be considered and future studies with prospective design are needed for more robust data collection. Finally, due to recruitment from a single ED clinic, the study included only treatment-seeking clinical patients. Further investigation is needed to characterize of people with ARFID in the community, which is useful to conclude the presentation of this study are unique to treatment-seeking population or can generalize to people with ARFID.
In conclusion, the present study highlighted differences in clinical features and personality between ARFID and RAN, as well as between pediatric and adult patients. Patients with ARFID showed less severe neuroticism compared to patients with RAN. Adult patients with ARFID exhibited more pronounced general psychopathology compared to pediatric patients, which suggest the need for age-specific treatment plans for ARFID. Recognizing these differences could inform more accurate assessments and personalized treatment plans for ARFID.
PDF Links
PubReader
ePub Link
XML Download
Full text via DOI
Download Citation
Print
