Systematic review of evidence for different treatment settings in anorexia nervosa

Table 1 Included randomised controlled trials evaluating the setting for treatment of anorexia nervosa

Ref.	Quality	Participants	Intervention	Outcome at end of treatment	Outcome at End of follow-up
Crisp et al[20],	Intention to treatment analysis, clear description of treatment and clear description of methodological problems High attrition rate, randomisation not described, lack of blinding, no power analysis and small numbers per group	90 adult female patients with DSM-IIIR AN with an illness duration of less than 10 yr. Specialist eating disorder service United Kingdom	Specialist, intensive inpatient treatment followed by 12 outpatient sessions 12 Outpatient individual and family psychotherapy including 4 sessions of dietary counselling 10 monthly outpatient group psychotherapy sessions (patient and parents) 4 sessions of dietary counselling Single assessment	Outcome took place 12 mo after the initial assessment regardless of when treatment finished All four groups increased weight significantly with weight gain in option 1 (9.6 kg), option 2 (9.0 kg), option 3 (10.1 kg) all being significantly higher than in option 4 (3.2 kg). All groups showed significant improvement in the Morgan-Russell scale. Groups 1 and 2 to P < 0.01 and groups 3 and 4 to P < 0.05 2 All groups showed significant improvement in the Morgan-Russell scale. Groups 1 and 2 to P < 0.01 and groups 3 and 4 to P < 0.05	2 yr follow up Only available for Individual outpatient and single assessment groups. Participants who received individual outpatient treatment had significantly higher weight, BMI, and mean matched population weight changes than the assessment only group. The treatment group experienced significantly greater improvement in the socioeconomic adjustment subscale or the Morgan Russell scale but there was not significant difference in other subscales or the global score of the Morgan-Russell scale
Kong[27]	Apriori power analysis and low attrition rate. Unclear blinding and concealment of randomisation. Small numbers per group and no intention to treat analysis	50 adult female participants with DSM-IV eating disorder. AN (32%), BN (42%) and EDNOS (19%). Specialist eating disorder service South Korea	Day treatment program. Multidisciplinary 4 d a week program for 8 to 14 wk Outpatient treatment (IPT, CBT) for 4 to 24 mo	Eating behaviours (binging and purging) EDI-2; weight and BMI; psychological symptoms EDI-2: depression (BDI) and self esteem (RSES). DTP patients compared to controls showed significantly less psychological symptoms of eating disorders (EDI-2), significantly less binge eating and purging, significantly higher mood and self esteem and significantly greater weight gain in AN patients	No follow up reported
Gowers et al[21]	Adequate allocation concealment with blinded assessors, apriori power analysis and adequately powered as a superiority trial with intention to treat analysis. High attrition rate	167 male and female adolescents (12 to 18 yr) with DSM-IV AN. England	Multi-disciplinary inpatient care in child and adolescent psychiatric units. ED pecialized outpatient care including motivational interviewing, CBT and family counselling Generic outpatient psychiatry treatment	Morgan Russell average outcome scale, HONOSCA (health of the nation outcome scale for Children and Adolescents)- clinician reported, EDI-2, HONOSCA self report, Family assessment device (FAD), Mood and Feeling questionnaire, BMI and Weigh for height	1 and 2 yr follow-up No significant differences in outcomes between the three interventions. Outpatient treatment more cost effective with higher treatment adherence. Increased parental satisfaction with specialist eating disorder treatment
Herpertz-Dahlmann et al[22]	Adequate allocation concealment with blinded assessors, apriori power analysis and adequately powered as a non-inferiority trial with intention to treat analysis. Low attrition rate	172, female patients aged 11 to 18 years with DSM-IV AN with a BMI below the 10th centile and their first admission to hospital for AN. The setting was 5 university hospitals and one general hospital for general child and adolescent psychiatry in Germany	3 wk inpatient admission followed by day patient treatment based on weight restoration, nutritional counselling, CBT and family therapy until patients maintained their target weight for 2 wk 3 wk inpatient admission followed by inpatient treatment based on weight restoration, nutritional counselling, CBT and family therapy until patients maintained their target weight for 2 wk	Primary outcome was the difference in BMI between admission and follow-up. Secondary outcome measures included: Morgan-Russell Outcome Scores, EDI-2, Brief Symptom inventory total scores, eating disorder readmission and the difference in cost between the two programs No significant difference in primary outcome. Insurance costs 20% less in day treatment	12 mo follow-up. No significant differences in primary and secondary outcome measures. Insurance costs 20% less in day treatment
Madden et al[34]	Adequate allocation concealment with blinded assessors, apriori power analysis and adequately powered as a superiority Low attrition rate	82 participants aged 12 to 18 yr, medically unstable with DSM-IV AN of less than 3 yr duration	Brief inpatient tabilizationn for medical tabilization (MS) (av. 22 d) followed by 20 sessions of outpatient family based treatment (FBT) Inpatient hospitalisation for weight restoration (WR) to 90% expected body weight (EBW) (38 d) followed by 20 sessions of outpatient manualised FBT	Outcomes at the end of treatment included the percentage of patients at full and partial remission, the percentage EBW and EDE global scores There were no significant differences between groups	12 mo follow up. Primary outcome was the number of days of hospitalisation following the initial admission Secondary outcome was the total hospital days used to and the percentage of patients at full and partial remission. Other outcomes included the percentage change in EDE global scores from baseline, readmission rates and the percentage of patients requiring treatment post protocol The only significant difference between groups was the total number of hospital days used which was significantly higher in the WR group

AN: Anorexia nervosa; FBT: Family based treatment; WR: Weight restoration; FAD: Family assessment device.