Sample Characteristics
There were 762 participants in the full sample, which included 576 individuals with current BN and 186 with current BED. Biological sex at birth determined by genotype was mostly female (87%), otherwise male (13%). Race was self-reported as 87% White, 6% mixed race, 3% Asian, 3% Black/African American, and 1% Native American. Ethnicity was reported as 90% non-Hispanic and 10% Hispanic. The mean ± SD for age was 30 ± 7 years. The mean number of binge-eating episodes over the past 28 days was 14.65 ± 9.61 and the mean global EDE-Q score was 4.05 ± 0.91, which are comparable to other acute samples in the literature (33, 34).
Treatment History
The sample was divided into two groups: a “treated group” (n = 471) and an “untreated group” (n = 291). Approximately 62% of the sample had received treatment for BN/BED at some point in their life. From most to least common, the treatments included psychiatric medication (45%, n = 341), CBT (39%, n = 294), IPT (26%, n = 196), inpatient care (8%, n = 59), residential care (8%, n = 58), and other psychotherapy (4%, n = 28). Some received multiple treatments. Approximately 11% received inpatient or residential care as their highest level of care (“inpatient group” n = 86), and 51% received outpatient care (i.e., CBT, IPT, other psychotherapy, psychiatric medication) (“outpatient group” n = 385). Emergency room care (4%) and weight-loss medication (9%) are reported here for descriptive purposes but were not used for classification into the treated group.
The untreated group included 214 (74%) individuals with BN and 77 (26%) with BED, and the treated group included 362 (77%) with BN and 109 (23%) with BED, with no significant difference in proportions, \(\chi\)2(1) = 1.07, p = .30. A lifetime diagnosis of AN was significantly associated with treatment history, with those having AN being more likely to receive treatment for BN/BED (\(\chi\)2(1) = 9.57, p = .002). Around 18% of the treated group and 10% of the untreated group had a history of AN, which was 15% of the sample. Because of this finding, sensitivity analyses were conducted by adding a history of AN as a covariate to the models to provide more context for interpreting the findings (see Supplement). The primary results are discussed by default, and differences are pointed out where applicable.
Sociodemographics
The untreated group had a significantly higher proportion of males and racial minorities than the treated group (Table 1). There were no significant differences on age or proportion of Hispanic participants. Although, a post hoc test among females indicated that Hispanic females were more likely to be untreated (45%) than non-Hispanic females (33%), \(\chi\)2(1) = 3.94, p = .04 and less likely to have accessed inpatient care (Fisher’s exact test p = .03). A post hoc test among males was not carried out because of low numbers.
There were significant differences by level of care on sex and race (FDR ps < .05). Compared with females, males were > 3 times more likely to be untreated than to have received inpatient care (OR = 5.65, 95% CI = 2.00, 16.13, FDR p = .004) or outpatient care (OR = 3.13, 95% CI = 1.98, 4.98, FDR p < .001). Compared with White participants, racial minorities were more likely to be untreated than to have received inpatient (OR = 3.39, 95% CI = 1.41, 8.13, FDR p = .01) or outpatient care (OR = 1.83, 95% CI = 1.20, 2.78, FDR p = .01). There were no significant differences in ethnicity or age across levels of care.
Current Clinical Features
Eating disorder psychopathology and behaviors. The untreated group had less severe current eating disorder symptoms, particularly compared to the inpatient group. EDE-Q weight concern was significantly higher in the treated group than the untreated group (FDR p = .007; Table 1), and this higher level of body dissatisfaction was not fully explained by BMI. Higher current BMI accounted for part of the variability in higher weight concern (p < .001), but the group difference remained (p = .007). Weight concern was significantly higher in both the inpatient and outpatient groups relative to the untreated group (FDR ps < 0.05). The other EDE-Q subscales and global score were not significantly different between the groups. The treated group had higher severity of vomiting and driven exercise symptoms. Compared with the untreated group, the inpatient group had 11 times higher odds of vomiting (OR = 11.17, 95% CI = 3.37, 37.05, FDR p < .001) and 2 times higher odds of driven exercise (OR = 2.19, 95% CI = 1.22, 3.91, FDR p = .02), Moreover, among those with positive counts of vomiting episodes, the outpatient group had more frequent vomiting than the untreated group (IRR = 1.80, 95% CI = 1.06, 3.03, FDR p = .04). The inpatient group was more likely than the outpatient group to engage in vomiting (OR = 8.74, 95% CI = 2.67, 28.59, FDR p = .001) and driven exercise (OR = 2.46, 95% CI = 1.39, 4.34, FDR p < .001). These differences on driven exercise were not significant in the model adjusting for a history of AN, suggesting that exercise for weight control was especially prevalent in participants with a history of AN (Table S2). There were no significant differences between the treated and untreated groups or the level of care groups in the odds of engaging in binge eating, fasting, or taking laxatives, compared to not engaging in these behaviors, or in the frequency of these behaviors among those who did engage in them (FDR ps < .05).
Comorbid mental health symptoms. Mental health comorbidity was prominent in the sample, with average scores corresponding to moderate depression (PHQ-9: M = 12.18, SD = 5.21) and mild-moderate anxiety (GAD-7: M = 9.89, SD = 5.37) according to the widely used PHQ-9 and GAD-7 thresholds. Additionally, 64%, 48%, and 49% of the overall sample met MDD, GAD, and ADHD screening criteria, respectively (24, 25). In general, the untreated group had significantly milder comorbid mental health symptoms. As shown in Table 1, the untreated group had lower depression, anxiety, and ADHD symptoms, and were less likely to screen positive for MDD, GAD, and ADHD, compared with the treated group in the uncorrected models; but after multiple testing correction, only ADHD symptom severity and screening positive for MDD were significantly different. The level of care analyses showed that as level of care increased, so did comorbid mental health severity. Both the inpatient and outpatient groups had significantly worse depression and ADHD comorbidity than the untreated group. Further, the inpatient group had worse anxiety, which may have been accounted for or at least was most commonly observed among those with a history of AN.
Comorbid DGBIs. DGBIs were common at the time of assessment, with 85% of the sample screening positive for any type of DGBI assessed. Unspecified functional bowel disorder (68%), IBS (31%), functional bloating (26%), and functional constipation (5%) were all relatively common. There were no significant differences between the untreated and treated groups on the prevalence of any of these diagnoses (FDR ps > 0.05). Uncorrected p showed a trend for a lower prevalence of unspecified functional bowel disorder in the untreated group (Table 2).
Current BMI. The mean current BMI of the sample was 32.67 kg/m2 (SD = 9.61). Untreated participants had a significantly lower BMI than treated participants, corresponding to a 1.6 kg/m2 difference. The untreated group had a significantly lower BMI than the outpatient group, corresponding to a 2.1 kg/m2 difference.
Lifetime Clinical History
History of eating disorder behaviors. The lifetime prevalence of behaviors used to control weight and shape in the overall sample was: fasting (80%), driven exercise (77%), diet pill use (56%), self-induced vomiting (55%), laxative use (40%), and diuretic use (27%), and only a minority (6%) reported never having used any weight control behaviors. The prevalence of behaviors to compensate for binge eating or overeating was also high. The most common was fasting (71%), and only 10% reported no use of any compensatory behavior. The mean age at onset of eating disorder behaviors in the sample was 15.71 years (SD = 5.60) and the mean age at onset of binge eating was somewhat higher at 17.82 years (SD = 6.74).
Differences were present on lifetime history of eating disorder behaviors between the untreated and treated groups. Untreated participants had an approximately two-year later onset of eating disorder behaviors than treated participants, F(1, 726) = 26.15, FDR p < 0.001). The untreated group was the least likely to ever use weight control behaviors such as vomiting, laxatives, or driven exercise for weight control, or compensatory behaviors. Compared with the untreated group, the inpatient group was 2–7 times more likely to report ever-use of vomiting (OR = 4.89, 95% CI = 2.74, 8.72, FDR p < .001), laxatives (OR = 2.03, 95% CI = 1.25, 3.31, FDR p = .01), and driven exercise for weight control (OR = 2.83, 95% CI = 1.39, 5.75, FDR p = .01) and vomiting (OR = 7.19, 95% CI = 4.02, 12.87, FDR p < .001) and laxative use (OR = 2.22, 95% CI = 1.35, 3.63, FDR p = .005) to compensate for binge eating. The outpatient group was 1.5–1.8 times more likely to report vomiting for weight control (OR = 1.59, 95% CI = 1.17, 2.16, FDR p = .008) and vomiting (OR = 1.84, 95% CI = 1.35, 2.51, FDR p < .001) and laxative use to compensate for binge eating (OR = 1.56, 95% CI = 1.12, 2.16, FDR p = .02) compared with the treated group. Some of these differences attenuated when adjusting for lifetime history of AN (Supplement).
Binge-eating duration was longer than a year for 69% of the sample. The duration was longer in the inpatient (OOR = 3.46, 95% CI = 1.79, 6.66, FDR p = .001) and outpatient groups (OOR = 1.57, 95% CI = 1.13, 2.20, FDR p = .02) than the untreated group, and the inpatient group had a longer duration than the untreated group (OOR = 2.19, 95% CI = 1.14, 4.20, FDR p = .03). There were no significant differences between any groups in the duration of vomiting, laxative, diuretic, diet pill, driven exercise or fasting symptoms (FDR ps > .05). There were several differences on symptom frequency. A higher frequency of symptoms such as binge eating, vomiting, laxatives, and driven exercise was observed in the inpatient group, followed generally by the outpatient group, followed lastly by the untreated group (pairwise FDR ps < .05). In comparisons between the untreated and the inpatient group, the OORs were 2.60 for frequency of binge eating (95% CI = 1.58, 4.41, FDR p < .001), 6.39 for vomiting (95% CI = 4.06, 10.04, FDR p < .001), 2.11 for laxatives (95% CI = 1.32, 3.37, FDR p = .005) and 2.68 for driven exercise (95% CI = 1.56, 4.46, FDR p = .001). For comparisons between the untreated and outpatient groups, the OOR for vomiting duration was 1.81 (95% CI = 1.35, 2.43, FDR p < .001. The group differences on frequency of laxative and excessive exercise use were not statistically significant in the model adjusting for lifetime history of AN, suggesting that they were more common in those with a lifetime history of AN, and perhaps accounted for by history of AN.
BMI history. The average lowest adult BMI was 23.50 kg/m2 (SD = 5.77) and the average highest was 35.62 kg/m2 (SD = 9.92). The inpatient group had a significantly lower illness-related lowest BMI than the outpatient and untreated groups (FDR ps < .05), but these differences were not significant in the sensitivity analysis after adjusting for lifetime AN. This indicates that those with BN/BED who were treated in inpatient/residential settings were also more likely to have had a history of AN. The outpatient group had a higher lifetime highest adult BMI than the untreated group (FDR p = .01). In the sensitivity analysis adjusting for history of AN, both inpatient and outpatient groups had higher highest adult BMIs than the untreated group (FDR ps < .05).