Background
A significant proportion of individuals with a diagnosis of anorexia nervosa (AN) describe the presence of an ‘anorexic voice’ (AV), an internal voice that tells them to restrict eating and lose weight. The AV is related to eating psychopathology, body mass index (BMI) and duration of the disorder. Moreover, some individuals experience their AV as another entity with which a relationship is formed. The relationship appears to evolve over time and whilst initially possessing positive qualities, it can become coercive and hostile, and can be a barrier to recovery (for a review, see Aya et al., Reference Aya, Ulusoy and Cardi2019).
People with AN often exhibit interpersonal difficulties in social relationships, with particular difficulties in asserting and expressing their needs and seeing themselves as inferior (Arcelus et al., Reference Arcelus, Haslam, Farrow and Mayer2013). Consistent with interpersonal theories, dysfunctional relating in social relationships often mirrors descriptions of the relationship with the AV. Worse eating psychopathology has been associated with individuals viewing themselves as inferior compared with their AV as well as relating in subordinate ways (Forsén Mantilla et al., Reference Forsén Mantilla, Clinton and Birgegård2018).
In the context of these relational challenges, addressing the individual’s interpersonal style with both the AV and other people may be a useful target for psychological therapies. Mixed evidence exists for oppositional and controlling approaches to relating to the AV. Alternatively, learning to respond assertively may be more appropriate and improve outcomes. An emphasis on assertive responding aligns with established eating disorder maintenance models, such as the cognitive-interpersonal model (Schmidt and Treasure, Reference Schmidt and Treasure2006). Relating Therapy (RT) aims to reduce the negative relating between the individual and both the AV and other people, and has generated favourable outcomes in the context of psychosis when offered over 16 sessions (Lincoln et al., Reference Lincoln, Schlier, Müller, Hayward, Fladung and Bergmann2024).
A case series of RT for the AV (n=3) found preliminary evidence that it was an acceptable intervention (Pugh et al., Reference Pugh, Bibby-Jones, Coyle-Stewart and Hayward2023). A subsequent consultation with people with lived experience suggested that RT should be offered over an extended duration to facilitate the guided application of learning to eating behaviours. The primary aim of this study was to explore the feasibility of evaluating an extended version of RT delivered over 24 sessions. A secondary aim focused on exploring signals of change.
Method
Study design and ethics
This was a mixed-methods feasibility study that used a case series with repeated within-person measures to explore signals of change. Case series data were collected during a baseline phase with three assessments (conducted weekly), followed by an intervention phase of RT delivered over a maximum of 24 sessions, with data collected throughout, and at 1-month follow-up. Optional exit interviews sought to explore participants’ experience of any changes during therapy. The study received NHS Health Research Authority (HRA) and Research Ethics Committee approval (South-central Berkshire REC; 22/SC/0413). A panel of people with lived experience of AN supported the study design and delivery.
Participants
The study aimed to recruit between 4 and 10 participants from the Eating Disorder Services within two NHS Trusts using an opportunity sampling method. Participants were required to be 18 years or older; have a clinician-reported primary diagnosis of AN; BMI of 18.5 or less; currently having experiences of a distressing AV; and willing and able to provide written, informed consent. A clinician-reported diagnosis of psychosis was an exclusion criterion. See extended report in the Supplementary material for details of all exclusion criteria
Intervention
RT supports individuals to respond assertively within difficult relationships with the AV and other people, and is conducted over three phases:
-
- Phase 1: socialisation to RT and its implications for the inter-relating between the individual and the AV/other people.
-
- Phase 2: exploration of themes within the relational history of the individual and their experience of difficult relationships.
-
- Phase 3: exploration and development of assertive approaches to the AV/other people. Experiential role-plays are used extensively to explore and practise relating in an assertive manner.
The intervention was delivered over of a maximum of 24 sessions by M.H. and L.S., with peer supervision facilitated by M.P. All participants were encouraged to engage in and continue with standard treatment as usual throughout the duration of the study (e.g. dietetic appointments).
Measures
Feasibility measures
Feasibility was evaluated in relation to: (1) the number of referrals received; (2) the proportion of referred patients found to be eligible; (3) the proportion of consented participants who were retained within the study and offered full datasets; (4) the proportion of non-missing data for each variable; and (5) the proportion of consented participants who reached the point of therapy exposure (a minimum of 16 sessions).
Safety was carefully monitored through the recording of any unfavourable sign or symptoms that developed or worsened during the study, whether or not it was considered to be related to the study intervention.
Measures for exploring signals of change – qualitative
The participants’ experience of change was explored through the analysis of qualitative data generated using the Change Interview to guide an optional exit interview.
Measures for exploring signals of change – quantitative
Sessional measures. Distress related to the AV was measured over the past week on an analogue scale of 0 (‘not at all’) to 10 (‘extremely’). Data regarding weight and BMI were collected during routine physical monitoring. A personal goal for therapy was identified and measured on a 11-point analogue scale over the past week. The subjective sense of separation between self and the AV was assessed on an analogue scale ranging from 0 (completely overlaps) to 10 (completely separate).
Four-weekly measures. Data relating to eating disorder psychopathology (Eating Disorder Examination Questionnaire Short; EDE-QS), depression (Patient Health Questionnaire; PHQ-9), anxiety (Generalised Anxiety Disorder Questionnaire; GAD-7), the power differential between participant and the AV (Voice Power Differential Scale), the relationship with the AV (The Experience of an Anorexic VoicE Questionnaire; EAVE-Q), and the participant’s style of relating to the AV (Approve-Voices) and other people (Approve-Social) were assessed using measures with established psychometric properties.
See the extended report in the Supplementary material for further details of the measures.
Qualitative data and data from four-weekly measures were collected by researchers who were independent of the delivery of RT. Data from sessional measures were collected by therapists.
Data analysis
Quantitative
Feasibility measures were summarised using descriptive statistics and compared against progression criteria, as appropriate. For case series data, at the group level, the unstandardised changes in mean scores from baseline (defined as the median of the three baseline scores) to post-intervention (final score) were calculated for AV distress, weight/BMI and eating disorder psychopathology for all participants with available data. Corresponding standardised effect sizes (Cohen’s d) were calculated by dividing the unstandardised change by the score’s baseline standard deviation and are reported descriptively. At the individual level, data for all measures were visualised in graphs and analysed for patterns of change within and between the baseline and intervention phases. Please see the extended report in the Supplementary material for details of the analysis plan and outputs.
There was no hypothesis testing and Cohen’s d effect sizes were calculated to support descriptive interpretation of findings and to identify preliminary signals of change that may inform outcome selection, measurement timing and study design in future trials.
Qualitative
Interviews were transcribed verbatim and analysed by C.C. and P.H. using reflexive thematic analysis.
Results
Participants
Eleven individuals were referred to the study between January 2023 and July 2023, and nine were eligible and provided consent. Eight participants commenced the intervention and seven completed the therapy by attending at least 16 sessions.
All participants who commenced the intervention identified as female; their mean age was 32 years (range 20–53; SD 11.96); most identified as White British; and many were in a long-term relationship. See the extended report in the Supplementary material for full demographic details.
Feasibility outcomes
Referrals and eligibility
The referral of 11 individuals exceeded the recruitment target. As nine (82%) of these individuals were eligible and consented, the target for eligibility (75%) was also exceeded.
Retention and missing data
Of the eight participants who commenced RT, six attended all of the four-weekly assessments that were offered. Overall, there was a high completion of data (86%, relative to a target of 80%). All participants who commenced intervention sessions completed more than 75% of measures (relative to a target of 75%). There was a decreased level of completeness for weight data (60%).
Therapy exposure
Seven participants reached therapy exposure (78% relative to a target of 75%) by attending at least 16 sessions, with those participants attending between 22 and 24 sessions. One participant withdrew before the first session due to starting a new job and another discontinued the intervention at session 14 due to anxiety about within-session role-plays.
Safety
Six adverse events were recorded for three participants (each reporting two adverse events). Two events related to a hospital admission whilst the remaining events related to self-harming behaviour and suicidal ideation. All were assessed by an independent rater as serious adverse events, but none was deemed to be related to the study or intervention.
Signals of change – qualitative
Exit interviews with six participants lasted between 56 and 68 minutes. Three superordinate themes emerged from the data analysis (Change is hard; RT helped me to see things differently; and Discovering the benefits of being different) and highlighted the dynamic process of the participant’s experience of change over the course of RT. Further details of the themes can be found in the extended report (Supplementary material).
Signals of change – quantitative
Eating disorder attitudes and behaviours
At the group level, there was an overall reduction in mean scores from baseline to post-intervention of –6.3 (95% confidence interval (CI): –12.7, 0.1; d=1.6). See Figure 1 for details of individual scores at each time point.
Change in scores on the EDE-QS.
AV distress
At the group level, there was an overall reduction in mean distress from baseline to post-intervention of –2.4 (95% CI: –4.9, 0.1; Cohen’s d=–1.8).
Weight and BMI
At the group level, there was an overall increase in mean BMI from baseline of 1.4 (95% CI: –1.0, 3.8; d=0.98) and an increase in mean weight of 4.5 (95% CI: –2.4, 11.4; d=1.4).
Full details of the data for weight, BMI, EDE-QS and all other measures can be found in the extended report (Supplementary material).
Discussion
This was the first study to explore the feasibility of evaluating RT when delivered over an extended duration of 24 sessions.
All feasibility outcomes were positively confirmed, except those relating to the data completeness. The study received sufficient and appropriate referrals, and most participants completed the intervention and schedule of assessments. However, the higher proportions of missing data observed for weight and BMI suggest that the successful collection of weight-related data may only be modestly successful within larger trials. Moreover, caution using weight or BMI as outcomes has been recommended, as consensus is lacking on how to meaningfully interpret change in weight-related outcomes (Austin et al., Reference Austin, Silva, Ilesanmi, Likitabhorn, Miller and Fialho2023). A good outcome should not solely be judged on the basis of weight restoration.
The exploratory data relating to signals of change offered encouragement as most participants reported a greater ability to engage with important and meaningful aspects of their lives, a re-defined relationship with the AV, and reduced distress around eating and body image.
At this preliminary stage only limited and tentative clinical implications can be drawn from the findings. Firstly, there may be therapeutic benefits to offering individuals with AN an opportunity to explore their experience of the AV. Secondly, RT may provide a specific framework for targeting some of the mechanisms that maintain distress related to the AV. Overall, the study findings offer a mandate for a feasibility randomised controlled trial comparing an extended course of RT with usual care.
This study had limitations: methodologically, the design did not allow causal conclusions to be drawn about the benefits of RT and findings may have been subject to the regression-to-the-mean phenomena; no time frame was specified a priori for participant recruitment; and the sample was small and relatively homogeneous in terms of ethnicity, sex, gender identity and education level. These issues need to be addressed in any future trials.
Supplementary material
To view supplementary material for this article, please visit https://doi.org/10.1017/S1352465826101386
Data availability statement
The data from this study will be made openly available via Figshare.
Acknowledgements
We are grateful to the members of the Lived Experience Advisory Panel who guided our journey through this study. Thanks are also due to the participants for helping us to learn from their experiences.
Author contributions
Callam Constant: Data curation-Lead, Formal analysis-Lead, Investigation-Lead, Methodology-Equal, Project administration-Equal, Writing - original draft-Equal; Lucy Serpell: Conceptualization-Supporting, Data curation-Equal, Investigation-Supporting, Project administration-Supporting, Writing - review & editing-Supporting; Matthew Pugh: Conceptualization-Equal, Supervision-Lead, Writing - review & editing-Supporting; John Fox: Conceptualization-Supporting, Writing - review & editing-Supporting; Tamara Leeuwerik: Supervision-Supporting, Writing - review & editing-Supporting; Anna-Marie Bibby-Jones: Formal analysis-Lead, Methodology-Supporting, Supervision-Supporting, Writing - review & editing-Equal; Penny Herbert: Formal analysis-Equal, Writing - review & editing-Supporting; Mark Hayward: Conceptualization-Equal, Data curation-Supporting, Formal analysis-Supporting, Funding acquisition-Lead, Investigation-Equal, Methodology-Supporting, Supervision-Equal, Writing - original draft-Equal.
Financial support
This study was supported by a grant from the ESRC Impact Acceleration Account at the University of Sussex.
Competing interests
The authors have nothing to declare.
Ethical standards
The study received NHS Health Research Authority (HRA) and Research Ethics Committee approval (South-central Berkshire REC; 22/SC/0413). All participants gave informed, written consent. Participants additionally gave informed, written consent for their anonymised data to be used within presentation and publications. All authors abided by the Ethical Principles of Psychologists and Code of Conduct as set out by the British Association of Counselling and Psychotherapy (BACP) and British Psychological Society (BPS).
Open access
Comments
No Comments have been published for this article.