A multi-centre cohort study of short term outcomes of hospital treatment for anorexia nervosa in the UK

The first aim of this paper is to describe the clinical features of patients admitted to a cross section of UK services for inpatient and day patient care on admission and at discharge. A second aim is to examine whether intra- (ED symptoms, mood, and motivation to change) and interpersonal factors (perceived expressed emotion, control, quality of social relationships) contribute to the response to inpatient care in terms of ED psychopathology. It is our expectation that ED symptoms, low mood and low motivation to change will contribute to a more negative response to inpatient care. Similarly, it is our expectation that high perceived carer expressed emotion and control and low quality of social relationships will contribute negatively to the response to inpatient care.

Participants were recruited from a consecutive series of inpatients and day patients admitted to specialist ED hospital services in the UK over a three year period (September 2008 and August 2011). Day patients in this context were patients who required non-residential intensive specialist treatment (≥ 4 days a week). The families of these patients agreed to participate in a randomised controlled trial (RCT) to examine the impact of a guided self-help skills training intervention on the outcomes of carers and the individual with AN post discharge from hospital treatment. However, only patient data are included in this paper. Trial protocol and recruitment procedure are described in more detail in another paper [27]. The inclusion criteria required individuals to be fluent in English, have a primary diagnosis of AN or Eating Disorder Not Otherwise Specified with Anorexic symptoms (diagnosed using clinical interview ICD 10 criteria), and have at least one carer consent to participate in the project. Some hospitals used a step down procedure whereby patients may be gradually moved to a less intensive form of treatment (i.e. from inpatient to day patient) prior to discharge into outpatient services or the community. Patients were considered discharged when they ceased to receive intensive treatment for their ED (that is, inpatient treatment or day patient treatment ≥ 4 days a week). Those who were transferred to residential care or day care ≥ 4 days a week where they continued to receive specialist and intensive ED treatment were not considered to be discharged; however, those who moved on to more generic supported housing services where they did not receive specialist intensive support were. Patients had to maintain their “discharged” status for a minimum of four weeks to be counted as formally discharged. If they returned to hospital within four weeks, this period was considered part of a continuous hospital admission. Details on whether patients completed the treatment programme or discharged themselves without recommendation by the treatment centre were not available for all sites and are therefore not included.

Data from participants of 14 hospitals were included in this paper. Twelve hospitals are adult ED specialist inpatient units and two are adolescent hospital units (one specialist ED and one general psychiatric with ED beds). Four sites recruited inpatients as well as day patients into the study. We wanted to have an ecologically relevant sample representative of current UK practice. Each service followed a different ethos with different practices relating to the transition back to the community. Since the numbers are small, data collected from day patients are presented together and are not separated by site. All hospitals are NHS operated except one which is private and accepts NHS referrals. More details can be found in a previous paper [27].

Ethics approval was granted by the Royal Free Hospital Ethics Committee (08/H0720/41) and site-specific ethics and governance approval was granted on all participating sites. This study was adopted by the Mental Health Research Network (MHRN) and Clinical Studies Officers supported the study.

Patients were assessed at admission and after discharge from hospital. Patients completed the first assessment on admission to hospital after providing consent for the study. Patients were not re-contacted by the study team until time of discharge. The hospital notified the treatment team that the patient had been discharged and questionnaires were subsequently sent to the patient at their home.

Body Mass Index (BMI): BMI (weight/height²) was obtained from the medical notes for each patient at admission and at discharge. This was available for all patients at admission. It was necessary to use self-reported weight for N = 14 patients (8%) at discharge.

Eating Disorder Examination – Questionnaire (EDEQ): A self-report measure assessing ED symptoms over the previous 28 days [28]. This instrument has good reliability and validity in ED samples [29]. High scores indicate greater ED psychopathology.

Depression, Anxiety and Stress Scale (DASS): A 21-item self-report measure to assess mood state over the past seven days using a 4-point Likert scale [30]. Total score as a measure of general distress or depression, anxiety and stress subscales can be used. High scores indicate higher symptomatology. This measure has good reliability and validity [30, 31].

Motivation to change: Patients are asked to rate, on Likert scales measuring 1 (not at all) to 10 (very), their importance and confidence in changing their ED behaviours.

World Health Organization – Quality of Life Questionnaire (brief version) (WHOQOL): Items are rated on a 5-point Likert scale pooled in four domains: physical health, psychological, social relationships and environment. Two single items evaluate the ‘Overall Quality of Life’ and ‘General Health’ facet. Good psychometric data have been reported for this scale [32]. High scores indicate better quality of life.

Levels of Expressed Emotion Scale (LEE): This is a 60-item self-report scale for patients to report on the presence of items representing high or low levels of expressed emotion in their carer using a true/false dichotomy. A total score as well as subscales can be derived [33, 34].

Psychological Control Scale (Youth self-report; PCS): This 8-item questionnaire assesses the level of perceived psychological control displayed by carers (reported for mother and father separately) [35].

All statistics were carried out using SPSS version 20.0.

Three participants (2 adult inpatients, 1 adolescent inpatient) were censored due to extremely long hospital admissions (duration stay > 132.5 weeks at time of writing). These patients remained in hospital at the time of data analysis and analyses were conducted without their discharge data.

The consort diagram of the patient and carer flow into the initial phase of the study is shown in Figure 1. In total, 30% of patients who were approached took part in the study. Reasons for not consenting into the study varied but the most common reasons were that patients were discharged prior to consent, didn’t have time to take part in research, or lacked a carer. Also, patients were excluded because their carers did not consent to take part in the study.

Inpatient treatment was successful for BMI (huge effect size). However, there was a large range in individual discharge BMI (11.9 – 24.4). The average rate of weekly weight gain over the duration of the admission varied widely between individuals (0.2 – 1.7 kg per week.) Effect sizes for change in BMI were large for all sites but there was a two-fold difference in the size of the effect between some sites. The rate of weight gain was lower than the minimum recommendation by NICE (0.5 kgs) in eight centres (seven adult).

The majority of adult inpatients remained in the AN BMI range at discharge from hospital (58%). Twenty (14%) of these had a discharge BMI of ≤ 15 and a further 62 (44%) had a BMI ≤ 17.5. Only 22 percent (N = 31) of patients had a BMI >19 at discharge.

Scores on EDEQ improved significantly with moderate sized effects. The only ED behaviour for which there was a significant reduction in the number of patients endorsing the behaviour was excessive exercise. There was a two-three fold difference in the size of effect between some centres.

Mean discharge scores for depression and anxiety remained in the severe range and ‘stress’ scores fell within the upper moderate range. There was no change in social wellbeing. Ratings on ‘importance to change’ significantly reduced over the treatment period and there were no changes in ‘confidence to change’.

There was a three-fold difference in the size of effects for mood and quality of life outcomes across centres. There was an increase in ‘importance to change’ in only two centres and an increase in ‘confidence to change’ in five centres.

The overall increase in BMI in adolescents was double that of the adult sample. The rate of weekly weight gain was slightly more than adults.

There were small reductions in ED specific psychopathology and, like adults, the only behavioural measure to show change was excessive exercise.

Effect sizes for improvements in mood were larger for adolescents than for adults and there was improvement in all WHO Quality of Life subscales. ‘Importance to change’ reduced but, unlike adults, ‘confidence to change’ increased over the treatment period.

There was a moderate increase in BMI but 40% of day patients remained in the AN weight range at discharge. The weekly rate of weight gain was modest with large individual variation (0.1kg; range: -0.2 – 1.2).

Change in ED specific psychopathology was significant (large effects) but 40% remained above the clinically significant threshold. There was no change in number of patients engaging in ED behaviours.

There were small to medium improvements in mood and quality of life. Both ‘importance to change’ and ‘confidence to change’ reduced over the treatment period.

The majority of patients had a severe and enduring form of illness and had failed to respond to earlier intensive treatment. They had high levels of functional impairment and most were living with their family of origin prior to their admission. Day patients had a higher weight at admission and less severe historical illness characteristics (lowest BMI, previous admissions) than inpatients. Adolescents had a higher BMI and less protracted illness than adults.

The change in clinical symptoms during the period of treatment was large for weight gain but modest for most other symptoms. The majority of patients were discharged when their BMI remained within the AN diagnostic weight range (≤ 17.5) and two thirds of the sample remained within the clinical range of general and specific psychopathology. Overall, the proportion of people engaging in compensatory behaviours did not show great change at discharge. Perceptions of quality of life and health improved but all domains remained low upon discharge relative to population norms [36]. However, this might be expected since patients need to reconnect socially and/or professionally having recently been discharged. Similarly, since psychotherapy sometimes needs a certain physical health to work in depth on psychological symptoms, it is possible that the lack of improvement in ED and mood symptoms is associated with the discharge weight of the group remaining within the pathological ED criteria. Patients’ weight and symptom status at discharge is of concern given previous reports that low weight predicts relapse [23, 37, 38].

BMI increase was smaller in day patients than in inpatients and at end of treatment, 40% of day patients remained within an anorexic BMI range. Day patients remained symptomatic at discharge and quality of life was poor. The change in BMI in the adolescent inpatient group was twice that attained for adults; this was associated with an elevated rate of weight gain and a small increase in the length of stay. There was also a larger improvement in mood and quality of life, and ‘confidence to change’ also increased over the treatment period which was not seen in the adults.

Clinical and social parameters predicted symptom change and some of these have the potential to be modified by change in clinical policy or practice. The level of patient’s mood psychopathology, confidence, social quality of life, perceived level of father’s control and carer expressed emotion predicted ED symptoms at discharge. The level of confidence has previously been found to predict the response to inpatient care [39] and has been found to have implications for post-inpatient outcomes [23]. Social function and family response to the illness (expressed emotion, father’s psychological control) predicted outcome and this supports the interpersonal maintenance model of EDs [24, 25]. Carer expressed emotion has previously been found to contribute to the outcome of adolescent AN [40‐42] and adults with bulimia nervosa (BN) admitted for inpatient care [43]. It is possible that if inpatient treatment targeted these psychosocial factors, and were inclusive of families, outcomes would be improved.

Inpatient care is able to produce an improvement in physical health but psychological and social health remains impaired. There are several ways in which treatment could be focused on those factors that might modify the response to inpatient care. Strategies to improve confidence to change such as motivational interviewing may be of value to improve outcome [44‐46]. Targeting interpersonal relationships with interventions that reduce expressed emotion [47, 48] may be of benefit. Strategies included in the recovery approach may reduce the social isolation [24, 49] and loneliness [50] that contribute to the poor social quality of life.

There are limitations of this study. Primarily, the data do not represent all patients admitted but only the group who consented to be involved in a study that involves their carers. It is possible that the sample is therefore biased towards including carers who were more actively involved in their loved one’s care. The samples recruited from each site or type of service were small, limiting the power to examine service level predictors. The percentage of adolescent inpatients and day patients is also small in comparison to the adult inpatient sample size which limits the overall outcome interpretation. All psychological outcomes were measured using self report questionnaires. Clinical interviews would improve reliability of findings. There were missing data for some of the predictor variables of interest, reducing data available for the regression analysis. However, it is a strength of the study that discharge data were collected from patients regardless of whether they completed the full inpatient/day patient programme, so it is representative of those who leave hospital treatment for AN. Data on whether patients took their own discharge, whether clinicians suggested an early discharge or whether they completed the programme were not assessed in this study since the information was not available for all sites and the definitions of each of these premature terminations of treatment are variable. It would be of interest to examine whether outcomes differed between these groups in future studies. Similarly, it would be interesting to examine outcomes based on different therapeutic approaches within and between sites. Lastly, we did not differentiate between patients who were admitted on a voluntary basis and those detained under the Mental Health Act. Despite these limitations, these data are a realistic reflection of current NHS practice of inpatient care in the UK and this paper presents a platform for future research in the area.