Patients with severe (BMI 15–15.99) or extreme (BMI < 15) AN [1] are often managed in an inpatient setting. However, some studies indicate that CBT-E can be suitable for these patients, provided that their medical conditions is stable [2]. The aim of this quality-assessment study was to determine the outcome of patients with severe and extreme AN in a real-world outpatient setting,


This quality-assessment study was performed at the department of eating disorders (DED) of the Psychiatric Clinic at Haukeland University Hospital, Bergen, Western Norway. The DED is a tertiary specialist ED unit that forms part of the public healthcare system in Norway.

The study sample comprised 21 consecutive patients aged between 17 and 51 years with severe or extreme AN (i.e. BMI < 16 kg/m2) enrolled between 1 January 2013 and 31 December 2016. Details of the patient flow to the DED are published elsewhere [3]. Figure 1 shows the recruitment and retention data for patients.

Fig. 1
figure 1

CONSORT diagram. BMI body mass index, CBT-E enhanced cognitive behaviour therapy for eating disorders, FBT family-based treatment, AN anorexia nervosa, DED department of eating disorders. Exclusion criteria: need of inpatient treatment at DED, healthcare was delivered at other unit, outpatient supportive care at DED because unable to perform psychotherapy or FBT at DED. Declined to participate; was offered outpatient CBT-E but declined to start (n = 9), other reason (n = 1)

EDs were diagnosed based on a clinical evaluation according to DSM-5 criteria [1]. BMI < 18.5, 15.0–15.99, and < 15.0 were applied as the inclusion criteria for AN, severe AN and extreme AN, respectively. The Mini International Neuropsychiatric Interview (version 6.0) [4] was used to screen for comorbid psychiatric disorders at baseline. At baseline the patients also completed the Norwegian versions of the Eating Disorder Examination Questionnaire version 6.0 [5] and the Clinical Impairment Assessment Questionnaire [6].

A physician assessed the patients before they received healthcare at the DED. Healthcare was overseen by an experienced medical specialist (S.F.).

CBT-E was performed as described in detail in the complete treatment manual [5] and in previous publications on DED treatment [3]. The patients did not receive any other ED psychotherapy while they were receiving CBT-E.


All statistical analyses were conducted using SPSS statistical package version 26.0 and R version 3.6.0 [7]. Wilcoxon rank sum test was conducted to compare the BMI at baseline, at the end of treatment (EOT) and at 1 year after the EOT. Cohen’s f effect sizes for within-sample changes in BMI from baseline to end of treatment were calculated, with values of above 0.4 representing large effect size [8].


The sociodemographic background and ED characteristics of the enrolled patients are presented in Table 1. The mean baseline BMI of the overall cohort was 14.8.

Table 1 Characteristics of 21 consecutive patients starting enhanced cognitive behaviour therapy (CBT-E) for anorexia nervosa

Nearly half of the 21 patients who started the treatment (n = 10, 47.6%) completed it (completers); the remaining 11 (52.4%) ended treatment prematurely (non-completers). Completers and non-completers were similar in age, BMI, number of years with an ED, rate of psychiatric comorbidity, number of previous ED treatment attempts and living situation (as listed in Table 1).

Outcomes among completers

The mean BMI values at baseline, EOT and 1 year after the EOT for the ten completers are shown in Table 2. There was a significant weight gain at EOT (Wilcoxon rank sum test: p = 0.0019), and the effect size for this change was large (Cohen’s f = 0.63).

Table 2 BMI of completers and non-completers

The percentage of completers achieving BMI ≥ 18.5 was 80% at EOT. The mean duration of CBT-E was 14.6 (SD 8.2) months (median, 13.2 months), and the mean number of sessions was 60.9 (SD 28.5) (median, 66.0).

Weight data 1 year after the EOT was available for all completers. There was no significant difference between BMI at EOT and BMI 1 year after the EOT for completers (Wilcoxon rank sum test: p = 0.59). At the 1-year follow-up, 80% of completers had BMI ≥ 18.5.

None of the completers entered inpatient care for ED in the first year after CBT-E.

Outcomes among non-completers

The BMI also increased significantly among the 11 non-completers as shown in Table 2. The percentage of non-completers presenting with BMI ≥ 18.5 at the last CBT-E session was 27% (3 of 11 patients). Follow-up took place at DED or in collaboration with DED (n = 5) or at district psychiatric centres in the Bergen area (n = 5). One of the non-completers was lost to follow-up. One year after premature end of CBT-E, the non-completers had a mean BMI of 18.01 (SD 3.4) (median, 19.15 kg/m2) and six of the eleven non-completers (55%) achieved BMI  ≥ 18.5 one year after premature end of treatment.

Thus, among the 21 patients with severe or extreme anorexia nervosa who started outpatient CBT-E 14 (66.7%) had BMI ≥ 18.5 at follow-up.


The present findings support previous studies indicating that CBT-E may be suitable for severe and extreme AN patients without acute medical complications. 20/21 (95%) of the patients who started outpatient CBT-E received follow-up healthcare at DED or at their local district psychiatric center. Among these 20 patients 14 (70%) had BMI > 18.5 kg/m2 one year after EOT. This quality-assessment study also supports previous findings indicating that BMI alone has limited value as a criterion for inpatient care. Further studies with larger samples are needed to evaluate the moderating effects of BMI and who is more likely not to complete treatment [9], and may need other forms of treatment [10].