© 2007 The Royal College of Psychiatrists
Economic evaluation of a randomised controlled trial for anorexia nervosa in adolescents
King’s College London, Centre for the Economics of Mental Health, Institute of Psychiatry, London
Biostatistics Group, Division of Epidemiology and Health Sciences, University of Manchester
Bolton Salford and Trafford Mental Health NHS Trust, Manchester
Peninsular College of Medicine and Dentistry, Exeter
University of Liverpool, Section of Adolescent Psychiatry, Academic Unit, Chester, UK
Correspondence: Sarah Byford, Box No 24, Centre for the Economics of Mental Health, Institute of Psychiatry, De Crespigny Park, London SE5 8AF, UK. Email: s.byford{at}iop.kcl.ac.uk
Declaration of interest None. Funding detailed in Acknowledgements.
See pp.
427–435, this
issue. 
 TOP ABSTRACT INTRODUCTIONMETHODRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
Background Young people with anorexia nervosa are often admitted to hospital for treatment. As well as being disruptive to school, family and social life, in-patient treatment is expensive, yet cost-effectiveness evidence is lacking.
Aims Cost-effectiveness analysis of three treatment strategies for adolescents with anorexia nervosa.
Method UK multicentre randomised, controlled trial comparing in-patient psychiatric treatment, specialist out-patient treatment and general out-patient treatment. Outcomes and costs assessed at baseline, 1 and 2 years.
Results There were 167 young people in the trial. There were no statistically significant differences in clinical outcome between the three groups at 2 years. The specialist out-patient group was less costly over the 2-year follow-up (mean total cost £26 738) than the in-patient (£34 531) and general out-patient treatment (£40 794) groups, but this result was not statistically significant. Exploration of the uncertainty associated with the costs and effects of the three treatments suggests that specialist out-patient treatment has the highest probability of being cost-effective.
Conclusions On the basis of cost-effectiveness, these results supportthe provision of specialist out-patient services for adolescents with anorexia nervosa.
|
TOPABSTRACTINTRODUCTIONMETHODRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
Anorexia nervosa is commonly associated with severe physical, psychological and social impairments, high levels of mortality (National Collaborating Centre for Mental Health, 2004) and a significant cost burden (Striegel-Moore et al, 2000; Simon et al, 2005). Young people with anorexia nervosa are often admitted to hospital for treatment. This is disruptive to school, family and social life, and in-patient treatment is an expensive option, yet evidence to support its cost-effectiveness is lacking (Romeo et al, 2005). A recent systematic review did not identify any economic evaluations of treatments for anorexia nervosa (National Collaborating Centre for Mental Health, 2004). One subsequent modelling study has been reported (Crow & Nyman, 2004). However, the authors acknowledge the limitations of modelling and the need for controlled trials. We report the results of an economic evaluation of psychiatric in-patient, specialist out-patient and general out-patient services for adolescents with anorexia nervosa carried out alongside a randomised controlled clinical evaluation – the Treatment Outcome for Child and adolescent Anorexia Nervosa (TOuCAN) trial (ISRCTN39345394).
|
TOPABSTRACTINTRODUCTIONMETHODRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
The aim of the TOuCAN trial was to explore the clinical and cost-effectiveness of in-patient, specialist out-patient and general out-patient services for adolescents with anorexia nervosa. The main economic hypotheses were that: (a) specialist out-patient services would be more cost-effective than general out-patient treatment in community child and adolescent mental health services (CAMHS); and (b) out-patient services would be more cost-effective than in-patient services.
Trial design
The design of this multicentre randomised controlled trial is described in
detail in the accompanying paper (Gowers
et al, 2007, this issue). In brief, adolescents aged
12–18 years with a DSM–IV diagnosis of anorexia nervosa were
recruited to the study from community CAMHS in the north-west of England
between March 2000 and December 2003. No exclusions were made on the grounds
of clinical severity, but the responsible clinician reserved the right to
refer for acute medical management according to urgent need. Participants with
severe intellectual difficulties or severe, chronic comorbid physical
conditions affecting digestion and/or metabolism were excluded. Randomisation
was carried out by an independent statistical centre. Stochastic minimisation
was used to control for gender, age (above and below 16 years) and body mass
index (BMI) (above and below 15.5).
Interventions
In-patient psychiatric treatment was for 6 weeks in the first instance,
extended as clinically indicated and determined by the treating service. The
in-patient services used a multidisciplinary psychiatric approach with the aim
of normalising eating, restoring healthy weight and facilitating psychological
(cognitive) change. Specialist out-patient treatment, described in detail
elsewhere (Gowers & Smyth,
2004; Gowers,
2006), was manualised and comprised individual
cognitive–behavioural therapy (CBT), parental counselling with the
participant, dietary therapy and multi-modal feedback. Treatment as usual in
community CAMHS generally involved a multidisciplinary, individual- and
family-based approach, with variable dietetic and paediatric liaison.
Clinical outcomes
Research assessors masked to treatment allocation carried out assessments
at baseline, 1 and 2 years after trial entry. The a priori primary
outcome measure for the clinical and economic evaluation was the
Morgan–Russell Average Outcome Scale (MRAOS;
Morgan & Hayward, 1988),
adjusted for adolescents. Full details of outcome measures, their reliability
and validity, are reported in the accompanying paper
(Gowers et al, 2007,
this issue).
Cost
The economic evaluation took a broad service-providing perspective,
including that of the health, social services, education, voluntary and
private sectors. Information on resource use was collected in interview at the
1- and 2-year follow-up assessments using the Child and Adolescent Service Use
Schedule (CA–SUS), developed by the authors in previous research with
young people and adapted for the purpose of the current study
(Byford et al, 1999;
Harrington et al,
2000; Barrett et al,
2006). Data on hospital contacts were collected from clinical
records to avoid patients revealing their treatment group to the research
assessors.
All unit costs were for the financial year 2003–2004. Costs in the second year were discounted at a rate of 3.5%, as recommended by the National Institute for Health and Clinical Excellence (National Institute for Clinical Excellence, 2004). This was varied from 0 to 6% in sensitivity analysis.
All National Health Service (NHS) hospital contacts, including the trial interventions, were costed using NHS reference costs (Department of Health, 2004). The unit costs of private sector in-patient stays were collected through direct personal communication with each facility. Unit costs of community health and social services were taken from national publications (Curtis & Netten, 2004). The costs of schooling came from a number of sources including various Ofsted reports (the inspectorate and regulatory body for schools in England, see http://www.ofsted.gov.uk) and published documents (Berridge et al, 2002; Independent Schools Council, 2004). Medications were costed using the British National Formulary (British Medical Association & Royal Pharmaceutical Society of Great Britain, 2004). Where necessary, unit costs were inflated to 2003–2004 costs using the Hospital and Community Health Services inflation indices (Curtis & Netten, 2004).
Statistical methods
All economic analyses were carried out on an intention-to-treat basis using
a statistical analysis plan drawn up prior to the analysis of the data. The
primary analysis was of total costs over 2 years for the sample of young
people with complete economic data.
Although costs were not normally distributed, analyses compared mean costs in the three groups using analysis of covariance with covariates for pre-specified baseline characteristics: site (Liverpool and Manchester), gender, age at baseline, baseline BMI and baseline MRAOS score. The robustness of the parametric tests was confirmed using bootstrapping (Efron & Tibshirani, 1993), as recommended by Barber & Thompson (1998). The impact of drop-out was assessed by comparing the baseline characteristics of participants who had missing data with those who had full economic data.
Cost-effectiveness was assessed through the calculation of incremental cost-effectiveness ratios (ICER) – the additional costs of one intervention compared with another divided by the additional effects of one intervention compared with another (Van Hout et al, 1994), in this case using the MRAOS measure of effectiveness. When more than two strategies are compared, ICERs are calculated using rules of dominance and extended dominance (Johannesson & Weinstein, 1993). Strategies are ranked by cost, from the least expensive to the most expensive, and if a strategy is more expensive and less effective than the previous strategy, it is said to be dominated and is excluded from the calculation of ICERs. This process compares strategies in terms of observed differences in costs and effects, regardless of the statistical significance of the difference.
Uncertainty around the cost and effectiveness estimates was represented by
cost-effectiveness acceptability curves
(Van Hout et al,
1994; Fenwick et al,
2001). Repeat re-sampling from the costs and effectiveness data
(bootstrapping) was used to generate a distribution of mean costs and effects
for the three treatments. These distributions were used to calculate the
probability that each of the treatments is the optimal choice, subject to a
range of possible maximum values (ceiling ratio, 
) that a
decision-maker might be willing to pay for a unit improvement in MRAOS score.
Cost-effectiveness acceptability curves are presented by plotting these
probabilities for a range of possible values of the ceiling ratio. These
curves incorporate the uncertainty that exists around the estimates of mean
costs and effects as a result of sampling variation and uncertainty regarding
the maximum cost-effectiveness ratio that a decision-maker would consider
acceptable (Fenwick & Byford,
2005).
Missing data were explored in three sensitivity analyses using the following data: (a) hospital cost data collected from clinical records and available for a larger sample of young people than full economic data from the CA–SUS; (b) hospital cost data collected from records plus missing non-hospital cost data imputed using the last value carried forward approach for participants with missing year-2 data; and (c) hospital cost data collected from records plus mean imputation by randomised group of missing non-hospital cost data. The results of all sensitivity analyses are reported in the data supplement to the online version of this paper.
|
TOPABSTRACTINTRODUCTIONMETHODRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
Participants
There were 167 young people entering the trial and they were randomised to in-patient care (n=57), specialist out-patient services (n=55) or treatment as usual by general CAMHS (n=55). Full economic data for the 2-year follow-up period were available for 135 young people (81%), 47 in the in-patient group, 45 in the specialist out-patient group and 43 in the general out-patient group. A comparison of baseline characteristics (site, age, gender, BMI and MRAOS score) revealed no significant differences between those included in the economic evaluation and those who were missing, and there was no difference overall in missing data between the three treatment groups. Length of follow-up varied somewhat (range 99–118 weeks), however there was no significant difference in length of follow-up between the three treatment groups (mean 105 weeks in the in-patient and general out-patient groups and 106 in the specialist out-patient group).
Outcomes
There were no significant differences between the three groups by intention
to treat at either 1- or 2-year follow-up on the MRAOS (in-patient 2-year
global score 8.3, specialist out-patient 8.4, general out-patient 8.3;
P=0.838). Full clinical out-come data are reported by Gowers et
al (2007, this
issue).
Resource use
Table 1 details the mean
number of contacts young people had with all services over the 2-year
follow-up period. Resource use differed little between the groups except for
in-patient and out-patient contacts. The general out-patient group spent more
time in hospital and had a greater number of out-patient attendances on
average than the specialist out-patient or in-patient groups. The specialist
out-patient group spent the least amount of time in hospital. Exploration of
hospital contacts over time reveals that a larger proportion of days were
spent in hospital in the first year (in-patient group 62 days, specialist
out-patient 35, general out-patient 65) than the second year (in-patient group
12 days, specialist out-patient 20, general out-patient 24). The in-patient
treatment of those allocated to out-patient treatment generally occurred after
assigned treatment had ended. Details of adherence to treatment are given in
the accompanying paper (Gowers et
al, 2007, this issue).
|
View this table: [in a new window] | Table 1 Use of resources during the 2-year follow-up period |
Hospital contacts reported in Table 1 include all specialties. However, the vast majority of contacts were psychiatric (71% of in-patient admissions and 90% of in-patient days) or paediatric (20% of admissions and 10% of in-patient days). Other specialties (9% of admissions and 0.2% of in-patient days) included gastroenterology, general medicine, haematology, intensive care unit, obstetrics, orthopaedics, plastic surgery and urology.
Time in education was similar across the three groups, however on average participants spent a significant proportion of the 2-year follow-up period out of education (approximately 10 out of the 24 months of follow-up).
Costs
Table 2 details the total
mean costs per participant over the 2-year follow-up period. There were no
statistically significant differences between the three groups. In terms of
observed differences, the specialist out-patient group was consistently
cheaper than the other two groups and the general out-patient group was the
most expensive of the three. The bootstrapped results differed little and are
thus not reported here. Hospital costs constitute the greatest proportion of
total costs (93% in each group), with few community health and social services
being used.
|
View this table: [in a new window] | Table 2 Total cost (£) per young person over the 2-year follow-up period |
Cost-effectiveness analysis
Using the rules of dominance described in the Method section, specialist
out-patient treatment (bootstrapped mean cost per participant £26 797;
bootstrapped mean effect 8.35) dominates the in-patient group (£34 371;
8.26) and the general out-patient group (£40 520; 8.26) since it is both
cheaper and more effective. Figure
1 illustrates the uncertainty associated with the costs and
effects of the three treatments at 2 years and demonstrates that if
decision-makers were willing to pay nothing for a unit increase in MRAOS
score, there is a 78% chance of specialist out-patient services being the most
cost-effective strategy, 16% for in-patient services and only 6% for general
CAMHS. The probability of specialist out-patient services being the most
cost-effective strategy decreases with increasing levels of willingness to pay
for gains in effectiveness, levelling out at around 47%, but remains higher
than the other two strategies over the full range of willingness to pay values
shown, and beyond.
 View larger version (12K): [in a new window] [as a PowerPoint slide] |
Fig. 1 Cost-effectiveness acceptability curve for Morgan–Russell Average
Outcome Scale (MRAOS) score.—, Specialist out-patient;—,
in-patient; - - -, general out-patient.
|
 View larger version (11K): [in a new window] [as a PowerPoint slide] |
Fig. 2 Cost-effectiveness acceptability curve for Morgan–Russell Average
Outcome Scale (MRAOS) score.—, Out-patient; - - -, in-patient.
|
|
TOPABSTRACTINTRODUCTIONMETHODRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
This paper reports the results of the first economic evaluation of alternative strategies for the treatment of anorexia nervosa using primary data collected from a randomised controlled trial.
Clinical outcomes, resource use and cost
There were no statistically significant differences in clinical outcomes
between the three groups, with results demonstrating similar improvements in
all groups over time. The specialist out-patient group was observed to be the
cheapest of the three groups and the general out-patient group the most
expensive. However, these differences were not statistically significant.
These findings were robust to changes in the discount rate and in analyses of
missing data.
Observed differences in total mean cost per participant were almost entirely due to differences in the length of time spent in hospital. Secondary healthcare costs accounted for over 90% of all costs and, of this, almost 90% was due to in-patient stays. The majority of in-patient stays took place in the first year. Although not randomised to psychiatric in-patient services, the general out-patient service group spent almost as much time in hospital as the in-patient group, suggesting that general CAMHS were less successful at maintaining these young people in the community than specialist out-patient services. With the exception of CAMHS, participants used very few community health and social services. Months in education was similar across the groups on average, but high-lighted the significant proportion of time participants spent out of education, presumably as a result of their illness.
At almost £17 000 per year on average, the annual service costs of caring for this group of young people were high. Although much higher than the cost of conditions generally treated in the community, for example conduct disorder with annual service cost estimates varying between £1300 and £3200 (Harrington et al, 2000; Romeo et al, 2006), this figure is similar to the cost of a cohort of young people treated in child and adolescent psychiatric in-patient wards, estimated to be £24 000 per admission (Green et al, 2007). Although slightly higher than the costs reported in this paper, the mean length of stay was longer at 116 days.
Cost-effectiveness
Specialist out-patient services were found to be the dominant treatment
option in terms of incremental cost-effectiveness (more effective and less
costly). Exploration of the associated uncertainty supported this finding. In
terms of our hypotheses, the data suggest that specialist out-patient services
have a higher probability of being cost-effective than general out-patient
services and that out-patient services (specialist combined with general) have
a higher probability of being cost-effective than psychiatric in-patient
services.
Limitations
Despite substantial differences in observed cost data, these differences
did not reach statistical significance. This may be due to inadequate sample
sizes for the economic evaluation. Sample size calculations were based on the
primary outcome measure, the MRAOS. Calculations on the basis of cost or
cost-effectiveness were not feasible at the design stage because of the lack
of any relevant published cost data. Although acknowledging this limitation,
the use of a decision-making approach to the economic evaluation provides
probabilistic evidence of the cost-effectiveness of the alternative treatment
strategies, given the data currently available. Although larger trials may be
considered in future research, this must be balanced against the cost of
additional research in a disease area where low prevalence rates necessitate
multicentre evaluation. Analysis of patients excluded because of missing
economic data did not suggest any bias; patients included in the economic
evaluation did not differ significantly from those excluded and there was no
evidence to suggest any bias in missing data between the three treatment
groups.
|
TOPABSTRACTINTRODUCTIONMETHODRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
This project is funded by the UK National Health Service Health Technology Assessment programme, Grant No. 97/42/02 and supported by teh North West MHEN. This paper is dedicated to Professor Richard Harrington.
|
TOPABSTRACTINTRODUCTIONMETHODRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
- Barber, J. A. & Thompson, S. G. (1998)
Analysis and interpretation of cost data in randomised control trials: review
of published studies. BMJ,
317, 1195
–2000.
[Abstract/Free Full Text] - Barrett, B., Byford, S., Chitsabesan, P., et al
(2006) Mental health provision for young offenders: service
use and cost. British Journal of Psychiatry,
188, 541
–546.
[Abstract/Free Full Text] - Berridge, D., Beecham, J., Brodie, I., et al (2002) Costs and Consequences of Services for Troubled Adolescents: an Exploratory, Analytic Study (Department of Health Report). University of Luton.
- British Medical Association & Royal Pharmaceutical Society of Great Britain (2004) British National Formulary 48. BMJ Books & Pharmaceutical Press.
- Byford, S., Harrington, R., Torgerson, D., et al
(1999) Cost-effectiveness analysis of a home-based social
work intervention for children and adolescents who have deliberately poisoned
themselves. Results of a randomised controlled trial. British
Journal Psychiatry, 174, 56
–62.
[Abstract/Free Full Text] - Crow, S. J. & Nyman, J. A. (2004) The cost-effectiveness of anorexia nervosa treatment. International Journal of Eating Disorders, 35, 155 –160.[CrossRef][Medline]
- Curtis, L. & Netten, A. (2004) Unit Costs of Health and Social Care. Personal Social Services Research Unit.
- Department of Health (2004) NHS Reference Costs. Department of Health.
- Efron, B. & Tibshirani, R. J. (1993) An Introduction to the Bootstrap. Chapman & Hall.
- Fenwick, E. & Byford, S. (2005) A guide to
cost-effectiveness acceptability curves. British Journal of
Psychiatry, 187, 106
–108.
[Abstract/Free Full Text] - Fenwick, E., Claxton, K. & Sculpher, M. (2001) Representing uncertainty: the role of cost-effectiveness acceptability curves. Health Economics, 10, 779 –787.[CrossRef][Medline]
- Gowers, S. G. (2006) Evidence based research in CBT with adolescent eating disorders. Child and Adolescent Mental Health, 11, 9 –12.[CrossRef]
- Gowers, S. & Smyth, B. (2004) The impact of a Motivational Assessment Interview on intitial response to treatment in adolescent anorexia nervosa. European Eating Disorders Review, 12, 87 –93.[CrossRef]
- Gowers S. G., Clark A., Roberts, C., et al
(2007) Clinical effectiveness of treatments for anorexia
nervosa in adolescents. Randomised controlled trial. British
Journal of Psychiatry, 191, 427
–435.
[Abstract/Free Full Text] - Green, J., Jacobs, B., Beecham, J., et al (2007) Inpatient treatment in child and adolescent psychiatry – an exploratory prospective study of health gain and costs. Journal of Child Psychology and Psychiatry, in press.
- Harrington, R., Peters, S., Green, J., et al
(2000) Randomised comparison of the effectiveness and costs
of community and hospital based mental health services for children with
behavioural disorders. BMJ,
321, 1047
–1050.
[Abstract/Free Full Text] - Independent Schools Council (2004) Annual Census 2004. Independent Schools Council.
- Johannesson, M. & Weinstein, M. C. (1993) On the decision rules of cost-effectiveness analysis. Journal of Health Economics, 12, 459 –467.[CrossRef][Medline]
- Morgan, H. G. & Hayward, A. E. (1988)
Clinical assessment of anorexia nervosa. The Morgan–Russell outcome
assessment schedule. British Journal of Psychiatry,
152, 367
–372.
[Abstract/Free Full Text] - National Collaborating Centre for Mental Health (2004) Core Interventions in the Treatment and Management of Anorexia Nervosa and Related Eating Disorders. British Psychological Society & Gaskell.
- National Institute for Clinical Excellence (2004) Guide to the Methods of Technology Appraisals. NICE.
- Romeo, R., Byford, S. & Knapp, M. (2005) Economic evaluations of child and adolescent mental health interventions: a systematic review. Journal of Child Psychology and Psychiatry, 46, 919 –930.[CrossRef][Medline]
- Romeo, R., Knapp, M. & Scott, S. (2006)
Economic cost of severe antisocial behaviour in children – and who pays
it. British Journal of Psychiatry,
188, 547
–553.
[Abstract/Free Full Text] - Simon, J., Schmidt, U. & Pilling, S. (2005) The health service use and cost of eating disorders. Psychological Medicine, 35, 1543 –1551.[CrossRef][Medline]
- Striegel-Moore, R. H., Leslie, D., Petrill, S. A., et al (2000) One-year use and cost of inpatient and outpatient services among female and male patients with an eating disorder: evidence from a national database of health insurance claims. International Journal of Eating Disorders, 27, 381 –389.[CrossRef][Medline]
- Van Hout, B. A., Al, M. J., Gordon, G. S., et al (1994) Costs, effects and c/e-ratios alongside a clinical trial. Health Economics, 3, 309 –319.[Medline]
Related articles in BJP:
- Clinical effectiveness of treatments for anorexia nervosa in adolescents: Randomised controlled trial
- SIMON G. GOWERS, ANDREW CLARK, CHRIS ROBERTS, ALISON GRIFFITHS, VANESSA EDWARDS, CLAUDINE BRYAN, NICOLA SMETHURST, SARAH BYFORD, and BARBARA BARRETT
BJP 2007 191: 427-435.[Abstract] [Full Text] - PETER TYRER
BJP 2007 191: 470.[Full Text]
This article has been cited by other articles:
 |
|
 |
|
  S G Gowers Management of eating disorders in children and adolescents Arch. Dis. Child., April 1, 2008; 93(4): 331 - 334. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. G. GOWERS, A. CLARK, C. ROBERTS, A. GRIFFITHS, V. EDWARDS, C. BRYAN, N. SMETHURST, S. BYFORD, and B. BARRETT Clinical effectiveness of treatments for anorexia nervosa in adolescents: Randomised controlled trial The British Journal of Psychiatry, November 1, 2007; 191(5): 427 - 435. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||