Introduction
Manual-based cognitive behavioural therapy (CBT; Fairburn, Marcus and Wilson, Reference Fairburn, Marcus, Wilson, Fairburn and Wilson1993) is the most extensively studied psychological intervention in the treatment of bulimia nervosa (BN) (Chakraborty and Basu, Reference Chakraborty and Basu2010). Cognitive behavioural treatment of BN (CBT-BN, Fairburn et al., Reference Fairburn, Marcus, Wilson, Fairburn and Wilson1993) has been shown to be more effective than other psychological treatments to which it has been compared, and to the use of pharmacological treatments (Wilfley and Cohen, Reference Wilfley and Cohen1997). In 2004, evidence-based guidelines proposed by the National Institute for Clinical Excellence (NICE, 2004) recommended CBT-BN as the treatment of choice for adults with BN and concluded that it should be routine practice in the National Health Service (NHS).
Previous research has shown that CBT-BN is associated with 40–50% of clients ceasing binge-purge behaviour (Wilson, Fairburn, Agras, Walsh and Kraemer, Reference Wilson, Fairburn, Agras, Walsh and Kraemer2002). With around half of clients not responding to CBT-BN, this led to the development of the transdiagnostic theory of eating disorders, extending its predecessor by proposing that although eating, shape and weight over-evaluation remain the “core psychopathology”, individuals may also experience additional maintaining processes.
In response, Fairburn, Cooper and Shafran (Reference Fairburn, Cooper and Shafran2003) derived an enhanced, transdiagnostic CBT (CBT-E); suitable for all clinical eating disorders including core psychopathology and additional maintenance mechanisms. The use of CBT-E for eating disorder psychopathology generally (regardless of DSM-IV diagnosis) rather than solely a treatment for BN is of clinical value as EDNOS is the most common category of eating disorder encountered in clinical settings (Fairburn and Bohn, Reference Fairburn and Bohn2005). Two versions of CBT-E exist; a “focused” version (CBT-Ef) exclusively addressing processes acting to directly maintain eating disorder psychopathology; and a “broad” version (CBT-Eb), also addressing one or more of the following additional maintaining processes; clinical perfectionism, core low self-esteem and interpersonal difficulties. The mechanism of mood intolerance was initially part of CBT-Eb; however, it was later moved to CBT-Ef (Fairburn, Reference Fairburn2008).
In an attempt to demonstrate the utility of CBT-E as a valid and appropriate treatment for BN and EDNOS, Fairburn et al. (Reference Fairburn, Cooper, Doll, O’Connor, Bohn and Hawker2009) conducted a randomized controlled trial (RCT) assessing the efficacy of CBT-Ef and CBT-Eb in outpatients with any form of eating disorder. The eligibility criteria stated that clients must have a BMI of 17.5 or above, be aged between 18–65 years, and have an eating disorder requiring treatment. One hundred and forty-nine patients were entered into the trial, receiving either CBT-Ef or CBT-Eb. Results revealed that of those who completed treatment, 66.4% had a global Eating Disorder Examination (EDE; Fairburn and Cooper, Reference Fairburn, Cooper, Fairburn and Wilson1993) score of less than one standard deviation above the community mean, indicating good outcome. Concerning those diagnosed with BN, 38.6% reported ceasing all binge-purge behaviour at the end of treatment, benefits that were maintained at a 60-week follow-up. Overall level of attrition was 22.1%.
In the full sample, there was no difference between the two versions of CBT-E (CBT-Eb and CBT-Ef), but for those clients with substantial additional psychopathology (of the type targeted by CBT-Eb), the use of CBT-Eb appeared to be more effective than CBT-Ef. This trial indicates that a single treatment type can benefit both BN and EDNOS diagnoses, supporting the transdiagnostic model and suggesting an increased utility over its predecessor CBT-BN, regarding the full range of patients treated.
Although RCTs are considered the gold-standard for inferring a cause and effect relationship, they are often criticized for not being relevant to the broad range of clients seen within a “typical” clinical setting. Fairburn et al.’s. (Reference Fairburn, Cooper, Doll, O’Connor, Bohn and Hawker2009) trial is a partial exception to this, due to the limited exclusion criteria stipulated and its focus on complex, additional psychopathology and maintenance mechanisms. However, it is still difficult to determine how such studies generalize to treatment conducted in a typical clinic, where controlled, prescriptive processes and treatment protocol may be difficult to adhere to, due to “real world” issues, including lengthy waiting-lists, limited resources and less intensive supervision arrangements.
An open trial conducted by Byrne, Fursland, Allen and Watson (Reference Byrne, Fursland, Allen and Watson2011) attempted to rectify this, by evaluating the generalizability of CBT-E in an outpatient clinic for adults with a full range of eating disorders in Western Australia. Byrne et al. (Reference Byrne, Fursland, Allen and Watson2011) found that of 176 referred clients, 125 (70%) entered the open trial and attrition rates were 40%. Of the 66 treatment completers, 56.1% were in predefined full remission, with 10.6% in partial remission; 66.7% of treatment completers had posttreatment global EDE-Q scores lower than one standard deviation above Australian community norms, indicating good outcome. Furthermore, significant improvements were revealed on all eating-related measures and associated psychopathology, such as depression, anxiety, stress, self-esteem and quality of life.
These findings suggest that CBT-E is generalizable to treatment conducted in a non-controlled clinical context, provided by therapists with a range of previous experience and training. Furthermore, as this study included individuals with the full range of eating disorders found in the community, this adds further evidence to the generalizability of CBT-E. However, the question remains whether CBT-E can be successfully delivered within the context of a UK NHS Eating Disorders Service (EDS), and, whether a service evaluation conducted within the UK can reproduce the promising findings demonstrated by Byrne et al. (Reference Byrne, Fursland, Allen and Watson2011).
Aims
This study analysed the impact of CBT-E within the Eating Disorders Service (EDS) of Cardiff and Vale Adult Mental Health Service. Specifically, this paper evaluated treatment outcomes for EDS clients who received CBT-E for treatment of BN or EDNOS between 2002 and 2011. Moreover, the evaluation aimed to investigate whether the response to CBT-E observed within this clinical context would be similar to those achieved in a RCT (Fairburn et al., Reference Fairburn, Cooper, Doll, O’Connor, Bohn and Hawker2009) and an Australian community study (Byrne et al., Reference Byrne, Fursland, Allen and Watson2011). Thus we were evaluating how well CBT-E implemented in a RCT context translated to the delivery of treatment within a UK community setting and hence whether CBT-E is an appropriate treatment in this context.
Method
Recruitment and inclusion for treatment
Clients were referred to the EDS primarily by Community Mental Health Teams (CMHTs). Appropriate referrals received an initial assessment, involving the completion of questionnaires and clinical interview with a clinical psychologist or specialist dietitian. Clients deemed suitable for treatment following assessment were placed on a waiting list until a treatment appointment became available. Eligibility criteria for accepted clients were: aged between 18–65 years; registered with a General Practitioner within Cardiff and Vale University Health Board catchment area; experiencing symptoms fulfilling the Diagnostic and Statistical Manual, 4th Edition (DSM-IV; American Psychiatric Association, 1994) criteria for BN or EDNOS for a minimum of 6 months; to be free from alcohol and/or drug addiction; and motivated to attend and engage in therapy sessions.
Treatment
EDS treatment was conducted on an outpatient basis and mirrored Fairburn, Cooper and Shafran's (Reference Fairburn, Cooper and Shafran2003) CBT-E. Although a standardized guide for treatment was not published until 2008 (Fairburn, Reference Fairburn2008), the two lead psychologists within the EDS underwent training in this treatment model in 2002 with Professor Fairburn and received year long, fortnightly group supervision, which continues to a lesser extent to the present day. Completion of the EDS treatment course was defined as successful progression through each of the four stages outlined by Fairburn (Reference Fairburn2008), but with some crucial differences within the treatment process:
1) Route into treatment included assessment appointments by two different professionals, one within the CMHT and one within the EDS. Patients meeting the criteria for this study were asked if they wished to proceed with this treatment and placed on a waiting list of 6–8 months, necessary due to limitations of dedicated staff time.
2) Patients opting into treatment were recommended a self-help book (Overcoming Binge Eating; Fairburn, Reference Fairburn1995) and invited to attend a lecture by the EDS assistant dietitian covering psycho-education of eating disorders, whilst on the waiting list. These options were put in place to increase patient motivation and engagement with the service whilst on the waiting list.
3) Sessions 0–8 were weekly, not twice weekly due to resource constraints.
4) Fairburn (Reference Fairburn2008) recommends the use of a preparatory session before commencement of treatment. Due to EDS staffing constraints and waiting list pressures, clients were not offered this.
5) Key stages 3 and 4 of treatment were supported with workbooks formulated by EDS psychologists following the training and supervision received from 2002 onwards from Professor Fairburn. The use of workbooks was discussed as a concept with Fairburn in supervision. Workbooks included shape and weight concern and checking, feeling fat, mindsets, dietary restraint and rules, and controlling eating. Workbooks also covered the impact of events, moods and eating, and finally “continuing your progress”. Workbooks summarized ideas learned from training with Fairburn prior to publication of the 2008 guide. They were designed to use as a guide with the client in session and to keep the therapist “on model”, along with supervision. Many therapists working in the EDS are seconded for half a day into the service on a training basis, with the rest of their time typically spent in a general adult mental health role. The workbooks represented a valued therapy aid for the service, allowing “generalists” to supplement their skills for this client group.
6) As the EDS is a clinical service, rather than a research trial, there was variance regarding the total number of sessions EDS clients received (delivered according to perceived need during treatment). The mean and median number of sessions was 20 and ranged from 6–40 sessions. The additional sessions accounted for in the higher range included work on significant early trauma, which would have been dealt with outside of the EDS in an RCT or where there was an alternative service to refer more complex cases. Where significant trauma was identified as a barrier to progress at session 6–8, CBT-E work was suspended and 10–20 sessions of trauma work offered before returning to complete stages 3 and 4 of the CBT-E. This occurred in fewer than 5% of cases.
Therapists
Within the EDS, clients were treated either by one of the two lead psychologists, who are eating disorder specialists trained and supervised by Professor Fairburn, or by a seconded non-specialist eating disorder therapist. These therapists were health professionals seconded to the EDS specifically to train in CBT-E and came from varying professional backgrounds, with differing experiences of psychological therapies. Seconded therapists were trained and supervised by the EDS lead psychologists. During their time at the EDS, seconded therapists worked with the EDS for half a day a week so they could see two cases at any one time, attend group supervision and additional training events. This commitment lasted for a minimum of one year. The limited time seconded therapists spent at the EDS was a factor in the restriction to once-weekly appointments in the first 8 weeks of treatment. Throughout the 2002–2011 period, the EDS experienced a high turnover of seconded therapists, with the number of therapists working within the service varying between 2 and 10 at any one time.
Ethics
The project was approved as a service evaluation and so no NHS Research Ethics Committee (REC) approval was required; however, university ethical approval was obtained.
Measures
Before commencing treatment, clients provided demographic data including: sex, age, marital status, ethnicity, occupation, age of onset of eating disorder, current BMI, and lowest ever weight. Clients were also asked to indicate their lifetime use of anti-depressants and contact with psychiatric services. Clients also completed pretreatment motivation scales.
Eating disorder features were assessed both prior to and following treatment, with the Eating Disorder Examination Questionnaire (EDE-Q; Fairburn and Cooper, Reference Fairburn, Cooper, Fairburn and Wilson1993), a 28-item self-report measure assessing the present state of an eating disorder. Eating disorder associated psychopathology was assessed with the Beck Anxiety Inventory (BAI; Beck, Epstein, Brown and Steer, Reference Beck, Epstein, Brown and Steer1988); and Beck Depression Inventory (BDI; Beck, Reference Beck1996).
Outcome variables
The main outcome involved changes in the EDE-Q, BDI and BAI for treatment completers, from pre to posttreatment. EDE-Q global scores generated from the four psychopathology subscales formed the second outcome measure. Normative comparisons were conducted to evaluate clinical significance of the intervention and offer indication of good outcome. Clinical significance was defined as posttreatment functioning falling within a normative range (Kendall, Marrs-Garcia, Nath and Sheldrick, Reference Kendall, Marrs-Garcia, Nath and Sheldrick1999). Posttreatment global EDE-Q scores were examined with scores less than one standard deviation above the UK global EDE-Q community mean (i.e. below 2.53) (EDE 17.0D; Fairburn, Cooper and O’Connor, Reference Fairburn, Cooper and O’Connor2014), indicative of good outcome. No follow-up data are available on the above outcomes.
Statistical analysis
Outcome data were analysed using both completer and intention to treat methods. For intention to treat analyses, pretreatment data were carried forward for those cases where end (posttreatment) scores were missing. The intention to treat sample consisted of all those who started treatment (n = 272); however, it excluded individuals who were currently receiving treatment (n = 26). Therefore the total number of individuals within the intention to treat sample was n = 246.
For comparisons between treatment starters and non-entrants, and completers and non-completers, categorical data were compared using chi-square tests, and continuous data compared with either Mann-Whitney U (for non-normal data) or independent t-tests (for normally distributed data). Pre and posttreatment data were compared with Wilcoxon signed-rank tests, due to the non-normal distribution of the data. Effect sizes were calculated using Pearson's r and Bonferroni correction was applied to control for multiple testing. Global EDE-Q scores were calculated for those who had complete data, pre- and posttreatment, for both completer and intention to treat analyses.
Results
Entry into EDS and patient characteristics
The EDS received 683 referrals between January 2002 and June 2011, with 272 patients commencing treatment. Figure 1 shows participant referral and flow through the EDS. “Non-entrants” were those who reached the top of the treatment waiting list and did not enter their first appointment, and those who attended their first appointment but decided not to engage with treatment. The 272 patients (60.3% of the 451 receiving an initial assessment) beginning treatment is larger than the 42.7% of patients entering Fairburn et al.'s (Reference Fairburn, Cooper, Doll, O’Connor, Bohn and Hawker2009) trial, but closer to the 70% of clients entering the Byrne et al. (Reference Byrne, Fursland, Allen and Watson2011) open trial from original referral and assessment. Characteristics of treatment starters are presented in Table 1.
Table 1. Characteristics of EDStreatment starters
Figure 1. Flow of patients through the EDS
Comparison of pretreatment measures and time waited between referral and initial assessment did not reveal any differences between treatment starters and non-entrants, following corrections for multiple testing. There were significantly more females within the treatment starters group (97.1% vs. 89.7%, p < .01); however, this did not remain significant, following corrections for multiple testing.
Attrition
Of the 272 treatment starters, 135 (49.6%) completed treatment; 88 (32.3%) dropped out of treatment; and 23 (8.5%) were categorized under “agreed closure” in which treatment was suspended on the joint decision of client and clinician, bringing the total “non-completers” to 111 (40.8%). A further 26 clients (9.6%) classified as “open” were still receiving treatment at the time of the study. Comparison of pretreatment measures between treatment “completers” and “non-completers” revealed significant differences for the BAI: t(240) = 3.517, p < .001), with pre-treatment BAI scores being lower for treatment completers. Completer and non-completer groups did not differ on any other pre-treatment measures or time waited between assessment and first treatment offer, following corrections for multiple testing.
Post-treatment CBT-E effects
For both treatment completers and the intention to treat sample, significant differences were found between all pre and posttreatment eating related psychopathology, associated psychopathology and for the following behavioural measures of the EDE-Q frequencies: binge episodes, binges, loss of control, vomiting and laxative use. Results are displayed in Table 2.
Table 2. Mean pre and posttreatment scores and effect sizes for treatment completers and the intention to treat sample on EDE-Q and associated psychopathology measures
Note: *Significant following bonferroni correction for multiple testing.
For treatment completers, 8 (7.5%) of the 106 patients for whom a global score could be calculated had a pretreatment score less than one standard deviation above the community mean (i.e. below 2.53). Posttreatment, this was the case for 83 (78.3%) treatment completers. Regarding the intention to treat sample, 12 (5.1%) of the 237 treatment starters for whom a global EDE-Q score could be calculated had a pretreatment score less than one standard deviation above the community mean, and posttreatment this was the case for 94 (39.7%) treatment starters. Fairburn et al.'s (Reference Fairburn, Cooper, Doll, O’Connor, Bohn and Hawker2009) RCT defined good outcome using global EDE scores, whereas the EDS utilized the EDE-Q, due to lack of feasibility to administer the EDE. Byrne et al. (Reference Byrne, Fursland, Allen and Watson2011) also employed the EDE-Q, from which global scores were calculated and compared with Australian community norms. Table 3 compares the three studies on this outcome variable.
Table 3. Comparison of the EDS results with those from the Fairburn et al. (Reference Fairburn, Cooper, Doll, O’Connor, Bohn and Hawker2009) RCT and Byrne et al. (Reference Byrne, Fursland, Allen and Watson2011) open trial on equivalent variables
Note: Good outcome was defined as having a posttreatment Global EDE-Q / EDE score less than 1 standard deviation above community norms.
Discussion
This evaluation aimed to explore the effectiveness of CBT-E as a treatment for BN and EDNOS offered by an Eating Disorders Service (EDS) within the Welsh NHS. Specifically, the evaluation aimed to investigate whether treatment outcomes associated with CBT-E offered within this clinical context would be similar to those achieved in an RCT (Fairburn et al., Reference Fairburn, Cooper, Doll, O’Connor, Bohn and Hawker2009) and an open trial (Byrne et al., Reference Byrne, Fursland, Allen and Watson2011).
The results demonstrated significant posttreatment improvements for treatment completers for eating psychopathology, measured by the EDE-Q, as well as for associated psychopathology assessed with the BAI and BDI, with medium to large effect sizes. Significant improvements were also demonstrated for the intention to treat sample; when including all treatment starters, however, they were associated with small to medium effect sizes. Similar findings in Byrne et al.'s (Reference Byrne, Fursland, Allen and Watson2011) open trial further indicated the successful impact and generalization of CBT-E to community, non-research settings.
Good outcome (global EDE-Q score of less than one standard deviation above the community mean) was achieved by 78.3% of EDS treatment completers, compared with 66.4% observed in Fairburn et al.'s (Reference Fairburn, Cooper, Doll, O’Connor, Bohn and Hawker2009) trial and 66.7% in Byrne et al.'s (Reference Byrne, Fursland, Allen and Watson2011) open trial, using comparable definitions. Intention to treat methods revealed that good outcome (defined above) was achieved by 39.7% of those individuals who started treatment, a rate lower than the 53% observed in Fairburn et al.'s (Reference Fairburn, Cooper, Doll, O’Connor, Bohn and Hawker2009) RCT and the 42.4% in Byrne et al.'s (Reference Byrne, Fursland, Allen and Watson2011) open trial. The higher rate of good outcome achieved by EDS treatment completers is interesting to note, especially considering a number of important service differences, potentially affecting treatment potency. Differing service variables are outlined in Table 4. These findings suggest that if the EDS is able to retain individuals within treatment, outcomes are positive, with a large proportion of treatment completers achieving good outcome, based on global EDE-Q scores. It therefore appears that a fundamental issue facing the EDS and potentially other non-research clinical services concerns retention of patients to treatment programmes.
Table 4. Service variables that may account for differences in the treatment outcome between the EDS, the RCT and the open trial
Reflective of this are the high attrition levels witnessed within the EDS (40.8%). EDS attrition rates were higher than those in Fairburn et al.'s (Reference Fairburn, Cooper, Doll, O’Connor, Bohn and Hawker2009) RCT (22.1%), but more comparable with the 40% observed within the Byrne et al. (Reference Byrne, Fursland, Allen and Watson2011) open trial and also within the 29–73% range of drop-out rates reported for eating disorder trials conducted on an outpatient basis (Fassino, Piero, Tomba and Abbate-Daga, Reference Fassino, Piero, Tomba and Abbate-Daga2009). Such elevated rates of attrition may be partially attributable to lower exclusion figures within the EDS than in Fairburn et al.'s (Reference Fairburn, Cooper, Doll, O’Connor, Bohn and Hawker2009) trial. For instance, of 449 patients assessed for treatment eligibility within the EDS 60.6% subsequently began treatment compared to 42.7% of those originally assessed within Fairburn et al.'s (Reference Fairburn, Cooper, Doll, O’Connor, Bohn and Hawker2009) RCT. Another possible explanation concerns the lack of preparatory sessions within the EDS. Fairburn (Reference Fairburn2008) emphasized the utility of such sessions to engage the client, formulate treatment expectations, and form a positive therapeutic relationship. They may also diminish barriers to treatment, reduce anxiety and promote retention within the treatment programme, potentially linking the association between attrition and higher pre-treatment anxiety rates. However, Byrne et al. (Reference Byrne, Fursland, Allen and Watson2011) did include preparatory sessions but still had a 40% attrition rate; hence further research may help clarify the role of preparatory sessions in this context. A third contributing factor may be the absence of initial twice-weekly sessions, which could have had a negative impact on initial engagement with the therapist and treatment.
Traditionally RCTs do not have waiting lists; however, increasing waiting lists are a notable concern within the NHS (Statistics for Wales, 2010) and may be relevant in explaining attrition rates in the eating disordered population, where motivation to change and to engage in treatment is notoriously low (Casasnovas et al., Reference Casasnovas, Fernández-Aranda, Granero, Krug, Jiménez-Murcia and Bulik2007). Within the EDS, there was no significant difference in terms of waiting times between completers and non-completers. It is however possible that waiting times contribute to higher attrition rates than those witnessed in an RCT context, especially given that Byrne et al. (Reference Byrne, Fursland, Allen and Watson2011) found longer waiting times for non-completers (25.9 weeks vs.18.15 weeks), and Carter et al. (Reference Carter, Pannekoek, Fursland, Allen, Lampard and Byrne2012) identified wait-list time as a significant predictor of dropout from the Australian clinic described in the open trial.
Potential explanations for the high attrition rates seen within the EDS highlight important issues and directions for service improvement as well as for the development of a more appropriately tailored service. This is especially relevant given the positive outcomes observed for those who remain in treatment and successfully complete within the EDS. In light of these findings, the EDS has since made service changes, including the introduction of a preparation session. Other discrepancies, however, cannot be so easily addressed within the NHS provision, such as specialist dedicated staff and shorter waiting list times.
Limitations of this study include the use of the self-report EDE-Q to derive outcome variables, as it is subject to both response and recall bias. However, validation studies have demonstrated high levels of agreement between the EDE-Q and the EDE in both the general population (Fairburn and Beglin, Reference Fairburn and Beglin1994) and in clinical samples (Carter, Aime and Mills, Reference Carter, Aime and Mills2001). In addition, the EDS did not include measures of additional psychopathology addressed by the “broad” version of CBT-E (CBT-Eb).The EDS could benefit from obtaining pre- and posttreatment measurements of such eating disorder features to assess whether CBT-Eb is effective in reducing their severity for those clients in which they are present.
Future research including follow-up assessments of treatment completers would be beneficial to assess the long-term utility of CBT. Eating disorders tend to run a chronic course with cycles of treatment and relapse episodes (Fairburn, Stice, Cooper et al., Reference Fairburn, Cooper and Shafran2003). It would therefore be interesting to explore predictors of both short and long-term response in order to identify within-client features associated with treatment efficacy. Research suggests early change to be a significant predictor of treatment outcome (Agras et al., Reference Agras, Crow, Halmi, Mitchell, Wilson and Kraemer2000). Therefore it may be advantageous to measure progress throughout treatment to identify those who will respond well to treatment and to provide further support to those not displaying such a positive early response.
In conclusion, this service evaluation provided support for the dissemination of CBT-E to treat BN and EDNOS. CBT-E was associated with significant improvements in eating disorder and associated psychopathology and improvements in the majority of behavioural symptoms for both completer and intention-to-treat methods. Moreover, global EDE-Q scores revealed high rates of “good outcome” for treatment completers, surpassing those found within the RCT (Fairburn et al., Reference Fairburn, Cooper, Doll, O’Connor, Bohn and Hawker2009) and open trial (Byrne et al., Reference Byrne, Fursland, Allen and Watson2011). Lower rates of good outcome observed for the intention to treat sample and high rates of attrition indicate the importance of retaining patients to treatment programmes.
Comments
No Comments have been published for this article.