To examine the costs and benefits of in-patient versus out-patient treatment for patients with eating disorders.

The authors searched MEDLINE (1966 to January 1999), PsycLIT (1967 to 1998), the Cochrane Library (Issue 4, 1998) and the Internet (including the National Library of Medicine and the NHS Centre for Reviews and Dissemination) using search strategies which are reported at length in the review. MEDLINE was searched separately for RCTs, case control studies, and case series studies using different search strategies. The authors also handsearched recent editions of European Eating Disorders Review and International Journal of Eating Disorders. References from review articles, RCTs, case-control studies and case series were checked for relevance to the review.

Study designs of evaluations included in the review

Included trials were:

1. Randomised controlled trials (RCTs) or case control studies comparing in-patient or residential care to out-patient or day- patient or GP care.

2. Case control studies comparing in-patient or residential care to out-patient or day-patient or GP care which included follow-up for in-patients and out-patients.

3. Case series of in-patient or out-patient care, irrespective of form of treatment given, starting with more than 100 patients and with a follow-up of 1 year or more.

Studies with fewer than 100 participants or with no details of mean length of follow-up were excluded.

Specific interventions included in the review

Treatment of eating disorders including weight stabilisation/restoration and attention to psychological factors (individual psychotherapy to remedy personal problems such as alienation and self-esteem and family therapy to reduce the impact on relatives and resolve particular relationship problems) and drug therapy.

Participants included in the review

Patients undergoing treatment for eating disorders (anorexia nervosa, bulimia nervosa, and binge eating disorder) on an in- patient or out-patient basis..

Outcomes assessed in the review

Deaths, global outcome categories (well versus remain ill) and Morgan-Russell mean scores for mean matched population weight (%MMPW), and global score.

How were decisions on the relevance of primary studies made?

The authors do not state how the papers were selected for the review, or how many of the reviewers performed the selection.

The authors assessed each of the study designs (RCTs, case control, and case series) using strategies appropriate to that design. For RCTs, the authors assessed method of randomisation and concealment of allocation, baseline characteristics and comparability, loss to follow-up, blinding, and whether intention to treat. For case control studies, the authors assessed whether the studies compared in-patient to out-patient treatment, whether the follow-up was for both in-patients and out-patients and whether conclusions matched the results. For case series, the authors assessed whether the case series matched the inclusion and exclusion criteria, if the case series was conducted prospectively, whether the method of selection of cases was identified and appropriate, and whether the duration and completeness of follow-up was reported and accurate. The authors do not state how the papers were assessed for validity, or how many of the reviewers performed the validity assessment.

Two reviewers extracted the data from all the included studies independently into pre-defined tables. Any discrepancies were resolved by discussion.

How were the studies combined?

The studies were combined in a narrative review of the results focusing on the outcome measures stated earlier.

How were differences between studies investigated?

The authors do not report a method for assessing differences between studies.

One unblinded RCT (90 participants), four case control studies (number of participants not stated), and seven follow-up case series studies (number of participants unclear) were included in the review.

Study quality:

The authors state that the RCT was not blinded and the difficulties in carrying it out means that the results must be viewed with caution. The evidence from the case series studies showed a wide variety of mortality and percentage well outcomes so that drawing conclusions from these studies as a whole is not possible. The evidence from the case-control studies found that while useful to show the differences between in-patients and out- patients at start of treatment, for physical and psychological factors, they do not show what subsequently happens following treatment. Study results:

The evidence from the single RCT with available data suggests that for the group of people with anorexia nervosa which is severe enough to consider in-patient care, but not severe enough for this to be essential, out-patient treatment is at least as effective as in-patient treatment. Further, the findings are suggestive of better outcomes for the out-patient groups. The benefits of out- patient and in-patient treatment appear to increase over time.

The results from the case-control studies and audit suggest that, in normal practice, those admitted as in-patients have, on average, a lower weight than those not admitted. There are few other differences between those normally treated as in-patients compared to those normally treated as out-patients, particularly for psychological factors.

The results from the case series show a wide variety of mortality and percentage well outcomes. This means that drawing conclusions from these case series as a whole is not possible.

The authors performed a telephone survey of the service use and costs of treating people with eating disorders in the West Midlands District Health Authorities and also designed a decision model to establish the overall effectiveness and cost effectiveness of the intervention. The intention was to estimate the cost utility (in, or translated into, the format of a multi- attribute utility scale, e.g. EuroQol EQ-5D) of in-patient treatment compared to out-patient treatment for eating disorders.

The available evidence shows no statistically significant differences in findings on outcomes between in-patient and out- patient care for people with eating disorders. This meant that the authors could not progress to an analysis of cost effectiveness.

The evidence on costs shows a wide variation in costs for in- patient episode (£4,349 to £32,636) and per out-patient treatment session (£34.70 to £68.44). It is very difficult to establish how accurate any of these methods were but, whatever method was used, out-patient treatment was always cheaper than in-patient treatment.

The authors state that more research needs to be undertaken on both in-patient and out-patient care for people with eating disorders. The lack of sufficiently powered RCTs and of good quality case series of in-patients and out-patients suggests that this treatment area has not been sufficiently researched in the past. If more people are to be treated solely in an out-patient setting in the future, it is vital that the progress of these people is followed carefully in order to confirm the suggested trends from the available research.

This review was well-conducted in general with only a few minor drawbacks. The authors have stated the research question and inclusion and exclusion criteria. The literature search appears to be thorough, although the searches were limited to English language publications and it is not clear whether there were searches for unpublished studies. It is possible that additional relevant studies were missed and there may, therefore, be publication bias.

The quality of the included studies was assessed using a scheme that took account of the different forms of studies included in the review and a discussion on the strength of the evidence produced by these different study designs. The authors have not reported how the articles were selected, or who performed the selection, or validity assessment. The authors did report who performed the data extraction and how this was performed.

The data extraction is reported in tables and discussed in the text of the review. The studies were not statistically combined and heterogeneity was not assessed however it was appropriate to combine these studies in a narrative review.

The authors conclusions appear to follow from the results but the authors state that these should be viewed with caution because of limitations in the included studies' designs and sample sizes.

Practice: The authors do not state any implications for practice.

Research: The authors state that much more research needs to be undertaken on both in-patient and out-patient care for people with eating disorders. If out-patient treatment for anorexia nervosa is to increase, it is vital that the progress of these people is followed carefully.

This is a critical abstract of a systematic review that meets the criteria for inclusion on DARE. Each critical abstract contains a brief summary of the review methods, results and conclusions followed by a detailed critical assessment on the reliability of the review and the conclusions drawn.