Participants

The sample consisted of 138 hospitalized underweight adult patients with ARFID or AN admitted to a specialized behavioral inpatient treatment program for eating disorders between 2003 and 2022. Patients with ARFID (n = 69) were matched to those with AN (n = 69) based on sex and age. A subsample of adults with ARFID (n = 27) who completed self-report questionnaires at admission were matched to 27 patients with AN based on sex and age (total sample with available self-report data N = 54). Sample demographics are presented in Table 1.

Table 1 Sample demographic and descriptive statistics characteristicsProcedures

Clinical and demographic data were collected as part of routine clinical care. A retrospective chart review approved by the Institutional Review Board of the Johns Hopkins University School of Medicine allowed abstraction of de-identified demographic and clinical data from the electronic medical record. This data included admission height and weight, pounds from target weight at admission, weight at discharge, length of stay (in days), achievement of target weight (yes or no), and reason for discharge (for clinical improvement vs not for clinical improvement [e.g., against medical advice, insurance, etc.]). Additional self-reported questionnaire data were collected from a subset of these patients who consented to participate in a longitudinal outcomes research study. Questionnaire data were collected using paper and pencil within the first week of admission. ED diagnoses were determined via the Structured Clinical Interview for the Diagnostic and Statistical Manual of Mental Disorders, 5th Edition (DSM-5; [7]). Diagnostic interviews were administered by postdoctoral fellows or trained research assistants supervised by a licensed clinical psychologist. Prior to the existence of DSM-5 and the introduction of the ARFID diagnosis, this study would have classified ARFID cases as Eating Disorder Not Otherwise Specified. After the release of DSM-5, all diagnoses were reviewed by a team of psychiatrists and clinical psychologists to ensure diagnoses consistently reflected the revised DSM-5 criteria. The SCID was administered by only one rater, and therefore inter-rater reliability is not available.

Treatment protocol

The inpatient eating disorders program employs a structured meal-based behavioral treatment protocol delivered within a multidisciplinary integrated, inpatient program. Primary treatment goals include rapid weight restoration for underweight patients and normalization of eating behavior. The nutritional protocol is 100% meal-based and nasogastric feeds are never employed. Patients consume three supervised meals a day in a group setting. Calories are advanced from 1200 to 2000 kcals/day (depending on admission BMI) to 3500–4000 kcals/day for individuals on weight gain protocol [11]. Calories above 2500 are administered via snacks and liquid supplements. With rare exceptions (e.g., patient was vegetarian at least three years prior to developing an ED, religious exceptions, or documented food allergy), food preferences or dislikes are not accommodated, as the goal is to help patients diversify their food intake. Patients may begin selecting menu items contingent upon completing 100% of meals served once at a calorie intake level of 3500 cal/day typically by day 10–12. Menu selections are reviewed by staff for compliance with the exchange plan. The nutritional rehabilitation protocol was identical for patients with ARFID and AN.

MeasuresWeight and BMI

Height and weight measured at admission and discharge were used to compute admission and discharge BMI. Individualized target weights set for each patient were a four-pound range (1.8 kg) based on the patient’s age, sex, and height centered on a BMI of 20.5 kg/m2 for patients over age 25. For patients aged 18–24, target weight was determined using growth charts when available. For patients whose baseline BMI was above the 50th percentile target BMI was set at the 50th BMI percentile. For those whose baseline BMI curve was below the 50th percentile but above the 25th, target BMI was calculated to fall on their premorbid BMI curve and for patients whose baseline BMI trajectory was below the 25th percentile, target BMI was set to the 25th BMI percentile. When growth charts were unavailable, target weight was set using the formula for adults over the age of 25 and adjusted by subtracting one pound (0.45 kg) per year of age below 25. Desired weight was assessed as a part of self-report questionnaires with one item (“How much would you like to weigh?”).

Eating disorder inventory-2 (EDI-2)

The EDI-2 is a 64-item self-report measure of cognitive and behavioral characteristics of weight and shape-related eating disorders [9]. Three subscales were used in the current study: Drive for Thinness, Bulimia, and Body Dissatisfaction. The EDI-2 has demonstrated good validity and reliability [9, 32] in patients with AN and BN, but no data are available on psychometric properties of the EDI-2 in patients with ARFID. Internal consistency for Cronbach’s alphas for Drive for Thinness, Bulimia, and Body Dissatisfaction were excellent in this study (α = 0.94, α = 0.83, and α = 0.91).

Eating disorder recovery self-efficacy questionnaire (EDSRQ)

The EDSRQ is a 23-item self-report measure of self-efficacy to refrain from acting on eating disordered behaviors and attitudes [21]. The Normative Eating Self-Efficacy subscale assesses confidence to eat without engaging in disordered eating behaviors (e.g., restricting, binge eating) and the Body Image Self-Efficacy subscale assesses confidence to maintain a realistic body image not dominated by pursuit of thinness. The EDRSQ has demonstrated good validity and reliability among individuals with weight and shape-related eating disorders [21, 24], but no psychometric data is available in ARFID samples. Internal consistencies for Normative Eating Self-efficacy and Body Image Self-efficacy were excellent in this study (α = 0.97, α = 0.93).

State and trait anxiety inventory (STAI)

The STAI is a 40-item self-report measure of anxiety as experienced in the moment (state subscale; STAI-S) and as a stable personality trait (trait subscale; STAI-T; [27]). The STAI has demonstrated good reliability and validity [12, 23]. Internal consistency for state and trait anxiety were excellent in the current study (α = 0.97, α = 0.94).

Beck depression inventory-II (BDI-II)

The BDI-II [2] is a 21-item self-report measure of depressive symptomatology. The BDI-II has strong psychometric properties, including internal consistency and factor validity [2, 28]. Internal consistency for the BDI-II in this study was excellent (α = 0.91).

NEO five-factor inventory (NEO-FFI)

The NEO-FFI [5] is a widely-used 60-item self-report questionnaire used to provide a concise measure of the big five personality factors including neuroticism, extraversion, openness to experience, agreeableness, and conscientiousness. The NEO-FFI has demonstrated adequate psychometric properties among individuals with weight and shape-related eating disorders [31]. Internal consistencies in the current study were excellent for neuroticism (α = 0.91), good for extraversion (α = 0.80) and conscientiousness (α = 0.87) and acceptable for openness to experience (α = 0.76) and agreeableness (α = 0.61).

Data analysisPower analysis

According to G*Power [6], a sample size of n = 64 in each group is required to detect a moderate effect size and a sample size of n = 26 is required to detect a large effect using an independent t-test analysis (power = 0.80, α = 0.05).

Treatment course

All analyses were conducted in SPSS v 28. The total sample (N = 138) was used for treatment course and outcomes analyses. Independent t-tests were conducted to compare patients with ARFID and AN on the following variables: Admission BMI, pounds from target weight at admission, length of stay, rate of weight gain, and discharge BMI. Pearson chi-square tests examined whether groups differed in whether they achieved target weight at discharge and reasons for discharge (clinical improvement vs not clinical improvement). Clinical improvement indicated that the patient had made sufficient progress in treatment and was ready to step down to a lower level of care for continued treatment and weight restoration. Not for clinical improvement indicated that the patient was discharged for other reasons including financial concerns, patient/family request (against medical advice), administrative discharge for non-compliance, transfer to medical or other psychiatric unit, or elopement.

Self-reported symptomatology

The subsample of patients who completed self-report questionnaires (N = 54) was used for these analyses. Independent t-tests were conducted to compare patients with ARFID and AN on the following admission variables: weight and shape-related ED symptoms, normative eating and body image self-efficacy, desired weight, anxiety, depression, and personality traits.