Adolescents (13–18 years old at baseline) in this study were referred from either the DBT service or MCCAED at the Maudsley Hospital. This case series reports on consecutive referrals between June 2017 and February 2020, the period in which the new adolescent adapted RO-A treatment programme was being delivered. All adolescents were screened for overcontrolled personality traits using the Assessing Styles of Coping Word-pair Checklist (ASC-WP)  followed by clinical interview assessing overcontrol factors such as risk aversion, perfectionism, emotional expressiveness, social connectedness, and rigid and rule governed behaviour.
Adolescents referred from the DBT service were all initially referred to standard DBT for treatment of repeated episodes of self-harm and low mood. If, during the initial assessment with the service, overcontrol was identified using the ASW-WP screening tool and clinical interview, RO-A was offered rather than standard DBT.
All adolescents referred from MCCAED were screened for overcontrolled using the same procedure (ASC-WP screen and clinical interview) if, after receiving family therapy for eating disorder (FT-ED), they continued to experience high levels of eating disorder behaviours and cognitions that interfered with daily functioning despite partial or full weight restoration. Persisting difficulties included, ongoing significant distress at mealtimes, significant cognitive rigidity and rules around food and eating, and/or significant social and education disruption due to these factors (e.g., missing school, struggling to socialise).
Exclusion criteria for this RO-A case series included psychosis, medical instability (see Junior MARSIPAN guidelines ), high psychiatric risk requiring inpatient treatment (e.g. imminent suicidal risk), emotional undercontrol and/or previous experience of RO DBT. No minimum weight was required for inclusion. See Fig. 1 for study flowchart.
Treatment intervention and model
Treatment in this study is an adolescent adaptation (RO-A) of the original RO DBT model described by Lynch . See above for more details on the modifications made. These changes were based on early feedback from adolescents that treatment length was too long and that materials were too adult focused.
RO-A includes 20 weekly 90-min skills class and a weekly 60-min individual session. Skills classes focus on teaching new skills to manage maladaptive overcontrol and includes mindfulness practice, homework provision and review. Skills classes consist of between two to eight individuals in treatment working together with one or two facilitators depending on the group size. The skills class focusses on teaching a range of skills designed to help adolescents express emotions more freely, engage in new novel behaviours, increase spontaneity and playfulness, live more flexibly, learn from feedback, strengthen social and community connectedness, and activate social safety systems. Individual sessions focus on applying these skills in the adolescent’s daily life, monitoring social signalling and overt overcontrolled behaviours, linking these with internal experiences and value-based goals. This includes the use of diary cards, in-session role plays and the use of chain analyses. See treatment manual for further details of treatment aims and structure [1, 16].
All adolescents were initially contracted to attend one full round of skills classes (n = 20) after which treatment was reviewed. Actual treatment length was based on individual goals and symptom presentation. Once adolescents had reached their identified value-based goals treatment ended, regardless of the number of individual or skills classes they had attended. Additional individual sessions and/or skills classes was offered if adolescents were actively working towards their value-based goals and using treatment effectively.
RO-A aims to reduce maladaptive overcontrol by targeting emotional expressiveness, cognitive flexibility, and social signalling. Improved social functioning and social signalling is hypothesised to lead to improved social connection, psychiatric symptom improvement and more global improvements in functioning.
All therapists involved in this study were employed by either the N&S CAMHS DBT service or MCCAED. RO DBT therapists represented the mix of professions present in both multidisciplinary teams, including clinical and counselling psychology, psychiatry, family therapy and nursing. All therapists attended 10-days of intensive RO DBT training delivered by approved RO DBT trainers, and attended weekly to fortnightly RO DBT consult with bi-monthly external supervision by a RO DBT approved supervisor.
Ethics approval and consent to participate
This study was approved by the South London and Maudsley (SLaM) CAMHS Service Evaluation and Audit Committee. As this study constitutes service evaluation or audit, NHS Research Ethics Committee approval was not required. SLaM CAMHS service evaluation and audit approval allows for analysis and publication of anonymised data extracted from case files without written consent from participants or carers. Outcome measures were administered as part of routine clinical care. All methods were performed in accordance with the stipulated guidelines and regulations.
Data collection and outcome measures
Outcome in this case series was measured as changes in overcontrol characteristics, relationship quality and psychiatric symptoms of depression and eating disorders. A range of self-report questionnaires were included that were selected to identify temperament, personality and coping factors associated with overcontrol in adolescents, as well as relationship quality and attachment. Validated adolescent measures were not available for the full range of overcontrol related factors as this is an emerging field. Adult measures were used in their absence. Symptoms of depression were also assessed using self-report measures to explore the relationship between changes in overcontrol factors and changes in psychiatric symptoms. Eating disorder symptoms were also assessed for those who reported eating concerns at assessment. Outcome measures were collected by clinical staff as part of routine clinical care.
Measures for screening and assessing overcontrol characteristics
The Assessing Styles of Coping Word-pair Checklist (ASC-WP)  was used as the initial screen for overcontrol. This 47-item self-report screening tool requires participants to choose one word from a pair of words that best describes them. Word pairs include one word that is more representative of over- and the other of undercontrol. The ASC-WP has not been validated with young people but was included due to an absence of any validated screening tools for overcontrol at the time of data collection.
The Emotion Regulation Questionnaire (ERQ)  is a validated 10-item self-report measure used to examine emotional regulation strategies via two subscales: cognitive reappraisal and the suppression of emotional expression. Cognitive reappraisal strategies refer to when someone changes their cognitions in order to change their emotional experience (example item: “when I want to feel less negative emotions [such as sadness or anger], I change what I’m thinking about”). The expressive suppression subscale assesses how much someone inhibits the behavioural expression of their emotions to regulate themselves (example item: “when I am feeling negative emotions, I make sure not to express them”). Cognitive reappraisal strategies are typically considered adaptive and associated with low psychological distress, whereas expressive suppression is considered less adaptive and associated with psychological distress and alexithymia . The ERQ has demonstrated good reliability and validity , good internal consistency , and has been used with adolescents . Internal consistency in the current study was good for the Reappraisal subscale (baseline a = .92; discharge a = .88), and moderate for the Suppression subscale (baseline a = .77; discharge a = .79).
The Negative Temperament subscale of the Schedule of Non-adaptive and Adaptive Personality for Youth (SNAPY-Y)  was included to assess level of maladaptive negative temperament and its stability across treatment. The subscale measures tendencies towards irritability, distress, fear, anger and sadness. The SNAP-Y has shown to be a valid measure of personality in adolescence that demonstrates good internal consistency, structural validity , and has available clinical norms [27, 28]. Internal consistency was moderate to good in the current study (baseline a = .78, discharge a = .81).
The Five Factor Obsessive Compulsive Inventory – Short Form (FFOCI) [29, 30] is a 48-item self-report assessment of risk aversion, cognitive flexibility, perfectionism, workaholism and punctiliousness. The FFOCI has not been validated for children and adolescents, but in the absence of a validated measure of obsessive-compulsive personality traits in children and adolescence, was included in this study. The FFOCI has demonstrated good discriminant validity and internal consistency with an undergraduate university sample . Internal consistency was variable in the current study and ranged from good to poor depending on the subscale (Risk Aversion baseline a = .68, discharge a = .61; Inflexibility baseline a = .59, discharge a = .81; Punctiliousness baseline a = .77, discharge a = .72; Perfectionism baseline a = .60, discharge a = .74; Workaholism baseline a = .86, discharge a = .89).
Reward processing was assessed using the Temporal Experience of Pleasure (TEPS) . The 18-item self-report measure assesses two aspects of trait-based reward processing based on Klein’s  model of anhedonia. Anticipatory pleasure (TEPS-ANT; “wanting”), the first subscale, examines the motivation for and expectation of pleasure and reward responsivity. The second subscale, consummatory pleasure (TEPS-CON; “liking”), measures the appreciation of positive stimuli and openness to different experience in the moment. Anticipatory, as opposed to the consummatory, aspects of reward processing have been associated with motivation, reinforcement learning and reward-based decision-making . The TEPS has not been validated with adolescents but has demonstrated good convergent and divergent validity, internal consistency and test-retest reliability in undergraduate university samples . Internal consistency within the current study was moderate to good (TEPS-ANT baseline a = .90, discharge a = .82; TEPS-CON baseline a = .69, discharge a = .84).
Measures assessing relationships quality
The Withdrawal subscale of the Youth Self-Report questionnaire (YSR-W)  is an 8-item self-report measure examining the degree of perceived social withdrawal and isolation. The YSR is a valid, reliable and frequently used measure to assess a range of problems in adolescents . Internal consistency was moderate to good in the current study (baseline a = .79, discharge a = .86).
The Social Connectedness Scale (SCS-R)  is a 20-item self-report measure used to assess connectedness that an individual feels in their social environment. Low scores are indicative of low levels of social connection. This measure shows good internal consistency and validity with an adult sample , however has not been validated with adolescents. Internal consistency was high in the current study (baseline a = .92, discharge a = .91).
The Attachment Styles Questionnaire (ASQ)  was used to define attachment characteristics and the quality of parental relationships. The ASQ consists of 40-items partitioned into five subscales including relationship confidence, need for approval, discomfort with closeness, pre-occupation and relationships as secondary. The ASQ has been shown to be valid and reliable, with good internal consistency [36,37,38] and has been used with adults and adolescents . Internal consistency ranged from good to poor in the current study, depending on the subscale (Confidence baseline a = .79, discharge a = .83; Discomfort baseline a = .86, discharge a = .84; Preoccupation baseline a = .70, discharge a = .70; Relationships as Secondary baseline a = .75, discharge a = .59; Need for Approval baseline a = .77, discharge a = .73).
Diagnostic assessment and measures of mental health symptoms
All adolescents in this case series completed the Development and Wellbeing Assessment (DAWBA) at assessment. The DAWBA is a widely used structured diagnostic assessment that generates DSM-5  and ICD-10  psychiatric diagnoses for two to 17-year olds . It has been shown to be a valid diagnostic tool  and may be more suitable than the widely used Eating Disorder Examination (EDE) diagnostic interview  for diagnosing adolescents with an eating disorder .
The Moods and Feelings Questionnaire (MFQ)  consists of 33-items used to screen for symptoms of depression in children and young adults. Scores of 27 and higher indicate the presence of depression [47, 48]. The MFQ was provided to all adolescents at baseline and discharge, regardless of symptom presentation. It has been shown to have good validity, reliability and internal consistency with adolescents . Internal consistency was good in the current study (baseline a = .92, discharge a = .91).
Incidence of self-harm was collected at baseline and discharge using a single-item questions. Adolescents self-reported whether or not they had engaged in any self-harm in the preceding 2 weeks.
The Eating Disorder Examination Questionnaire (EDE-Q, v6) was completed at baseline and discharge by those who reported eating concerns at assessment (n = 23). The EDE-Q is a 28-item measure with a total (global) score made up by four subscales: restraint, eating concerns, shape concerns and weight concerns. It has good internal consistency  and has been used previously with clinical  and community adolescent samples . Internal consistency was moderate to good, depending on the subscale (Global Score baseline a = .79, discharge a = .83; Dietary Restraint baseline a = .86, discharge a = .84; Eating Concerns baseline a = .70, discharge a = .70; Shape Concerns baseline a = .75, discharge a = .59; Weight Concerns baseline a = .77, discharge a = .73).
Percentage of median Body Mass Index (%mBMI) adjusting for age and gender (BMI/median BMI for age and gender × 100) was also recorded to assess changes in physical health for those who reported eating concerns. This is the recommended method for children and adolescents with anorexia nervosa . In this study, any young person under 90%mBMI was classified as underweight, and under 85%mBMI as significantly underweight.
The Shapiro-Wilkes test was used to test the distribution of the data. Paired t-tests were used for normally distributed data and Wilcoxon signed-rank test for non-normally distributed data to compare differences between baseline and discharge data. Cohen’s d was used to measure effect sizes for the paired t-tests (> 0.3 = small; > 0.5 = medium; > 0.7 = large). Non-parametric data effect size was estimated using r (> 0.1 = small; > 0.3 = medium, > 0.5 = large). McNamar’s test was used to compare rates of self-harm (present/absent) in the 2 weeks preceding assessment and the 2 weeks preceding discharge. Effect size was estimated using Cramer’s V. Internal consistency for each measure and subscale was assessed using Cronbach’s alpha. Pearson’s r correlations were conducted to explore the relationship between changes in overcontrol related factors and changes in symptoms of depression and eating disorders from baseline to discharge. Due to the exploratory nature and sample size, significance testing was not conducted, rather 95% confidence intervals are reported. All statistical analyses were performed using SPSS version 26.
To examine potential sampling bias in missing data at discharge, analyses were conducted to compare those who had paired data (completed assessment measures at both baseline and discharge) to those who did not across key demographic and clinical factors. Results showed that there were no differences between those with paired data compared to those without with regard to age, referral team (MCCAED or DBT service), primary diagnostic category (eating, mood or anxiety disorder diagnosis), severity of mood symptoms (MFQ at baseline) or the presence of self-harm. For the subgroup referred with eating concerns there was also no difference in weight (%mBMI) or severity of eating disorder psychopathology (EDEQ Global score) at baseline. Further analysis was conducted to examine difference in treatment characteristics. There was no difference between those with paired data and those without with regard to duration of treatment (in weeks), the number of skills classes attended, or the number of individual sessions attended.