AIM: To compare outcomes in anorexia nervosa (AN) in different treatment settings: inpatient, partial hospitalization and outpatient.

METHODS: Completed and published in the English language, randomized controlled trials comparing treatment in two or more settings or comparing different lengths of inpatient stay, were identified by database searches using terms “anorexia nervosa” and “treatment” dated to July 2014. Trials were assessed for risk of bias and quality according to the Cochrane handbook by two authors (Madden S and Hay P) Data were extracted on trial quality, participant features and setting, main outcomes and attrition.

RESULTS: Five studies were identified, two comparing inpatient treatment to outpatient treatment, one study comparing different lengths of inpatient treatment, one comparing inpatient treatment to day patient treatment and one comparing day patient treatment with outpatient treatment. There was no difference in treatment outcomes between the different treatment settings and different lengths of inpatient treatment. Both outpatient treatment and day patient treatment were significantly cheaper than inpatient treatment. Brief inpatient treatment followed by evidence based outpatient care was also cheaper than prolonged inpatient care for weight normalization also followed by evidence based outpatient care.

CONCLUSION: There is preliminary support for AN treatment in less restrictive settings but more research is needed to identify the optimum treatment setting for anorexia nervosa.

Anorexia nervosa (AN) is characterized by the restriction of energy intake relative to requirements leading to a significantly low weight for gender, age, developmental trajectory and height. Individuals with AN maintain an intense fear of gaining weight or becoming overweight even though they are underweight. They have a disturbance in the way in which they perceive their body weight and shape, there is undue influence of weight and shape on self evaluation and/or they deny the seriousness of their illness[1].

Lifetime prevalence rates for AN range from 0.9%-2.3% with its onset primarily in adolescence[2-5]. AN is the third most common chronic disorder affecting adolescent females, with and average mortality rate of 5%, up to 18 times greater than in non-affected women aged 15-24 years[6]. The course of the illness is often protracted with less than 50% of individuals with AN achieving full recovery in a period of seven years after the onset of their illness[7]. Physical complications include growth retardation, osteoporosis, infertility and changes in brain structure[8], while psychological complications include depression, anxiety disorders including obsessive compulsive disorder and cognitive impairment[9-11].

While the complications of AN have been well documented, there is limited evidence to guide as to what is the most appropriate treatment setting. Current expert consensus and guidelines[12,13] support outpatient care as the first line treatment in AN and there is a growing body of evidence to guide such care, particularly the role of family treatment in adolescents[14]. There are, however, few studies comparing different treatment settings in AN and few studies into the role of inpatient treatment in AN. Hospitalization of patients with AN for the management of acute medical instability (e.g., hypothermia, hypotension, electrolyte abnormalities and cardiac arrhythmias) is thought to be essential in preventing mortality associated with AN[8,15], however, the benefits of inpatient weight restoration and the assumption that hospital is the best venue for refeeding once medical stability has been achieved remain unsupported by current evidence[16]. This problem is highlighted by a recent meta-analysis of 57 clinical trials in AN that found no difference in treatment effects based on the style of therapy or treatment setting[17].

Treatment settings for AN include inpatient, partial hospitalization (also known as day hospital care) and outpatient. Inpatient treatment programs are usually multidisciplinary and treatment focuses on weight restoration, normalizing eating behavior and facilitating psychological change through nutritional education, supervised meals and individual and group psychotherapy[18,19]. In published trials admissions have been prolonged with a length of stay ranging from several months to 15.2 wk[22-24]. Partial hospitalization or day programs are similar to inpatient treatment in terms of a multidisciplinary approach, treatments offered and intensity and duration of treatment but with no overnight stays[22]. They are potentially more flexible and thus amenable to matching the individual level of motivation for change with goals of treatment[24]. Outpatient treatment generally involves once or twice a week sessions with a therapist of a single discipline and may be either individual or family based[12,13,25]. Outpatient treatments provide care in the least restrictive setting and thus are favored by current psychiatric practice[12,13].

The present article aims to assess the effects of different treatment settings including inpatient, partial hospitalization and outpatient on treatment outcomes in AN and to assess the effects of continuing hospitalization for weight normalization compared with brief hospitalization for medical stabilization in AN. Primary outcome measures included measures of clinical improvement including weight as measured by body mass index (BMI) in adults or percentage expected body weight in children and adolescents (%EBW) and changes in eating disorder psychopathology.

Of the five trials only one demonstrated differences in treatment outcomes for AN in different treatment settings, notably better weight and psychological outcomes for those treated with partial hospitalization compared to outpatient treatment[27]. This trial, however, had a moderate risk of bias. While the trial had a low attrition rate, subject numbers were low (n = 50) and less than one third of the patients had AN (n = 14) with the remainder meeting diagnostic criteria for bulimia nervosa or eating disorder not otherwise specified. There was no intention to treat analysis and it was unclear if the outcome assessor carrying out the statistical analysis was blind to patient allocation.

None of the other four trials showed differences in treatment outcomes for AN in adults or adolescents based on treatment settings. The treatment studies did demonstrate reduced costs for shorter inpatient hospital admissions compared with longer inpatient hospitalization[34] and reduced costs for partial hospitalization[22] and outpatient treatment[35] compared to inpatient care. On the basis of treatment outcome it is not possible to recommend one treatment setting over another, however, based on the consistent finding of increased costs it is difficult to support the use of inpatient treatment in AN in the absence of medical instability or other risk issues and even in these circumstances difficult to support inpatient treatment beyond medical stabilization[15,17,34].

These four trials were of moderate to high quality with the risk of bias minimized through the use of intention to treat analysis in all of these trials and in the three most recent trials, adequate allocation concealment, blinded assessors and apriori power analysis[21,22,34]. The number of trials is, however, small and compares different aged populations including adults[20] and adolescents[21,22,34], different treatment setting comparisons with only the one trial comparing partial hospitalization with inpatient treatment being adequately powered as a non-inferiority trial[22].

Strengths of this present review included the use of two authors in the literature search and evaluation of articles and data extraction. The main limitations were the inclusion of English language and use of published papers only.

It is clear is that whilst all of these trials are supportive of treatment in less intensive and restrictive settings more research is needed comparing both different treatment settings and different lengths of inpatient stays before conclusive recommendations can be made regarding the optimum treatment setting in AN. This is particularly the case in adults with AN where studies have been of lower quality. It is also important that when new studies are planned that they are adequately powered as non-inferiority trials.