Introduction
Anorexia nervosa (AN) is associated with substantial impairments in functioning across multiple life domains and has among the highest mortality rates of any psychiatric disorder, a risk that increases with commonly co-occurring psychiatric disorders such as depression (Ágh et al., Reference Ágh, Kovács, Supina, Pawaskar, Herman, Vokó and Sheehan2016; Søeby et al., Reference Søeby, Gribsholt, Clausen and Richelsen2024). Positive treatment outcomes can be elusive, particularly for adults. First-line evidence-based treatments for adult AN include enhanced/eating disorder-focused cognitive behavioural therapy (CBT-E/CBT-ED), the Maudsley model of Anorexia Nervosa Treatment for Adults (MANTRA), and Specialist Supportive Clinical Management (SSCM) (National Institute for Health and Care Excellence, 2004). However, approximately 50% of patients who start these treatments will discontinue early and approximately 50% of patients who complete treatment do so without meaningful reductions in their cognitive and/or behavioural symptoms (Solmi et al., Reference Solmi, Wade, Byrne, Del Giovane, Fairburn, Ostinelli, De Crescenzo, Johnson, Schmidt, Treasure, Favaro, Zipfel and Cipriani2021; van den Berg et al., Reference van den Berg, Houtzager, de Vos, Daeman, Katsaragaki, Karyotaki, Cuijipers and Dekker2019).
Improved outcomes for adult AN may come from enhancements in how available therapies are used or matched to patients; the development of new AN-focused therapies; or the adaptation of existing therapies to AN. Schema therapy is one example of a therapy that may suit adaptation to AN. Schema therapy developed out of CBT but incorporates theoretical, formulation and practice elements from interpersonal, psychodynamic, attachment and Gestalt therapies (Young et al., Reference Young, Klosko and Weishaar2003). There is a particular emphasis on experiential techniques and the therapeutic relationship. Schema therapy was developed for individuals with long-standing psychological difficulties that had not responded to traditional CBT and is increasingly recognised as an evidence-based intervention for individuals diagnosed with borderline personality disorder (BPD)/emotionally unstable personality disorder (Arntz et al., Reference Arntz, Jacob, Lee, Brand-de Wilde, Fassbinder, Harper and Farrell2022; Sempertegui et al., Reference Sempértegui, Karreman, Arntz and Bekker2013). In the UK, the National Institute for Health and Care Excellence (NICE) and NHS England list schema therapy as one of several psychological therapies that may be helpful for individuals diagnosed with BPD or ‘complex emotional needs’ (National Institute for Health and Care Excellence, 2009; NHS England, Reference NHS2024).
We hypothesise that schema therapy may offer benefits for adults with AN who have not benefited from first-line therapies. Over the remainder of this manuscript, we propose a model of schema informed CBT for AN and subsequently test this model via a case series (n=11). The following sections of the Introduction provide an overview of schema therapy; outline the existing evidence base for schema therapy with eating disorders (EDs); and introduce our proposed schema informed CBT model for AN. We subsequently present the case series methodology and results.
Schema therapy and schema-informed CBT
Core to the schema therapy model is an emphasis on unmet emotional needs in early life, which give rise to ‘early maladaptive schemas’ (EMS). EMS are broad, pervasive views of oneself, others and the world that are unhelpful in nature and shape interpretation of events and interactions (Young et al., Reference Young, Klosko and Weishaar2003). Schemas consist of thoughts, behaviours and emotions in addition to memories and bodily sensations. When activated, EMS contribute to difficult emotional states. ‘Schema modes’ describe these momentary emotional states associated with an EMS or the activation of multiple EMS, and associated coping responses (Arntz and Jacob, Reference Arntz and Jacob2013). Schema therapy aims to help patients connect with and meet their core emotional needs (‘Vulnerable Child’ mode) and build a ‘Healthy Adult’ state that can coordinate healthy coping responses. There is a focus on challenging overly critical or demanding schema modes and moderating dysregulated/impulsive/overcompensating coping modes. The therapeutic relationship is emphasised as a key enabler for this work (Arntz et al., Reference Arntz, Jacob, Lee, Brand-de Wilde, Fassbinder, Harper and Farrell2022; Young et al., Reference Young, Klosko and Weishaar2003).
Schema therapy as originally developed is long-term in nature, spanning 18 months to 3 years or more. This limits accessibility and scalability. More recently, and with service constraints in mind, schema-informed CBT has been proposed for cases underpinned by pervasive and chronic unmet emotional need. This time-limited model anchors treatment on a schema mode formulation and uses change technique from both CBT and schema therapy (Kiers and de Haan, Reference Kiers and de Haan2024; Moorey et al., Reference Moorey, Byrne and Ruths2020). This way of working may expand the accessibility of schema therapy and improve treatment outcomes for patients who have not benefited from first-line CBT interventions. Briefer versions of schema therapy (defined here as up to 1-year of out-patient individual and/or group sessions) have been used successfully with anxiety disorders (Gude and Hoffart, Reference Gude and Hoffart2008; Hoffart et al., Reference Hoffart, Versland and Sexton2002; Kiers and de Haan, Reference Kiers and de Haan2024), major depressive disorder (Kiers and de Haan, Reference Kiers and de Haan2024) and substance misuse disorders (Ball and Young, Reference Ball and Young2000; Oraki, Reference Oraki2019), as well as personality disorders (Farrell et al., Reference Farrell, Shaw and Webber2009; Kiers and de Haan, Reference Kiers and de Haan2024). However, the quality of these studies has varied (e.g. see Masley et al., Reference Masley, Gillanders, Simpson and Taylor2012; Peeters et al., Reference Peeters, van Passel and Krans2021) and active control groups have not always been included. A distinction has been made between brief applications of schema therapy that primarily focus on cognitive techniques (schema-informed CBT) and those that primarily focus on experiential techniques (van Vreeswijk and Broersen, 2013 [Dutch], as cited in Kiers and de Haan, Reference Kiers and de Haan2024). To date, studies have not elucidated which cognitive, behavioural and experiential techniques are most effective in brief applications of schema therapy. Support does exist for imagery rescripting (one experiential technique) as a standalone treatment for post-traumatic stress disorder (PTSD) and other mental health presentations characterised by aversive memories (e.g. Boterhoven de Haan et al., Reference Boterhoven de Haan, C. W., Fassbinder, Van Es, Menninga, Meewisse and Arntz2020; Kip et al., Reference Kip, Schoppe, Arntz and Morina2023). Imagery rescripting for PTSD may also lead to changes in schema coping modes even when a schema mode formulation is not used (Daniëls et al., Reference Daniëls, Meewisse, Nugter, Rameckers, Fassbinder and Arntz2025).
Schema therapy and eating disorders
Studies have identified a greater number of EMS in participants with EDs compared with healthy controls and dieters, even after controlling for depressive symptoms and self-esteem; and a greater presence of schema coping modes in participants with EDs compared with healthy controls and other patient groups (Maher et al., Reference Maher, Cason, Huckstepp, Stallman, Kannis-Dymand, Millear and Allen2022; Marney et al., Reference Marney, Reid and Wright2024; Pugh, Reference Pugh2015; Simpson and Smith, Reference Simpson and Smith2019). EMS have also been found to predict the severity of ED symptoms and to mediate associations between childhood sexual abuse and EDs (Fasolato et al., Reference Fasolato, De Felice, Barbui, Bertani, Bonora, Castellazzi and Bonetto2024; Jenkins et al., Reference Jenkins, Meyer and Blissett2013).
There appear to be some diagnostic differences in EMS and schema modes, although evidence is mixed and still emerging (Bär et al., Reference Bär, Bär, Rijkeboer and Lobbestael2023; Mitchell et al., Reference Mitchell, Huckstepp, Allen, Louis, Anijärv and Hermens2024; Luck et al., Reference Luck, Waller, Meyer, Ussher and Lacey2006; Pugh, Reference Pugh2015). Restrictive low weight EDs may be associated with primary avoidance of affect, i.e. the avoidance of affect prior to schema activation, whereas binge/purge behaviours may function as secondary avoidance strategies, i.e. reducing affect after schema activation (Luck et al., Reference Luck, Waller, Meyer, Ussher and Lacey2006; Waller et al., Reference Waller, Corstorphine and Mountford2007b). In the binge/purge subtype of AN, where low weight, restriction and binge/purge behaviours are present, there may be particularly high rates of early maladaptive schemas and schema coping modes (Pugh, Reference Pugh2015).
Given the above findings, it is unsurprising that schema concepts are already thought about in some versions of CBT-ED. For example, Waller et al. (Reference Waller, Cordery, Corstorphine, Hinrichsen, Lawson, Mountford and Russell2007a) suggest that schema-focused methods be considered when EMS and associated painful emotions may be maintaining ED symptoms. Here, the focus is primarily on core beliefs and cognitive strategies, including thought diaries, flashcards, positive data logs and schema ‘dialogue’. Imagery rescripting is also proposed (Waller et al., Reference Waller, Cordery, Corstorphine, Hinrichsen, Lawson, Mountford and Russell2007a). Similarly, core beliefs and imagery rescripting are considered in the protocol for 10-session CBT for EDs (CBT-T; Waller et al., Reference Waller, Turner, Tatham, Mountford and Wade2019). In these models, the core formulation remains cognitive-behavioural in nature, with difficult early experiences and EMS positioned alongside over-evaluation of eating, weight and shape and their control (the ‘core psychopathology’ in CBT-ED) and associated vicious cycles between restrictive eating, binge eating and/or purging. In line with the evidence for transdiagnostic CBT-ED (Fairburn, Reference Fairburn2008; Waller et al., Reference Waller, Cordery, Corstorphine, Hinrichsen, Lawson, Mountford and Russell2007a), these applications of schema therapy have been proposed across ED diagnoses.
There is also emerging evidence for schema therapy for EDs that is based on a schema, rather than CBT-ED, formulation. A 2023 systematic review identified four studies in this area and concluded that there is overall but preliminary evidence for the benefits of schema therapy for EDs (Joshua et al., Reference Joshua, Lewis, Kelty and Boer2023). The authors highlighted the limited literature available and variability in methodology and protocols. The reviewed studies included single case evaluations (Simpson, Reference Simpson, van Vreeswijk, Broersen and Nadort2012) and small group evaluations (n=8–10) of weekly or twice weekly schema therapy over 6 months (George et al., Reference George, Thornton, Touyz, Waller and Beumont2004; Simpson et al., Reference Simpson, Morrow, van Vreeswijk and Reid2010). There has also been one randomised controlled trial (n=112) of schema-focused CBT, standard CBT-ED, and appetite-focused CBT in the treatment of bulimia nervosa and binge eating disorder (McIntosh et al., Reference McIntosh, Jordan, Carter, Framptom, McKenzie, Latner and Joyce2016). The three trial groups each received weekly individual sessions for 6 months followed by monthly individual sessions for a further 6 months, and the three treatments were similarly effective in reducing binge eating and associated ED psychopathology. Large effect sizes were seen for all treatment conditions (McIntosh et al., Reference McIntosh, Jordan, Carter, Framptom, McKenzie, Latner and Joyce2016). Building on this early evidence, a proposed group treatment protocol has been developed (Calvert et al., Reference Calvert, Smith, Brockman and Simpson2018), in addition to a book-based guide for Schema Therapy for EDs (Simpson and Smith, Reference Simpson and Smith2019). A protocol has recently been published for a planned randomised controlled trial of group schema therapy for eating disorders relative to CBT-ED (Mares et al., Reference Mares, Roelofs, Zinzen, Béatse, Elgersma, Drost and van Elburg2024).
Since Joshua and colleagues’ review (2023), a series of studies from Iran have evaluated short applications of group schema therapy (12× 45-minute sessions; 20× 60-minute sessions; 20× 45-minute sessions) for participants with AN (Ansari et al., Reference Ansari, Asgari, Makvandi, Heidari and Seraj Khorrami2020), bulimia nervosa (BN) (Bagheri Sheykhangafshe et al., Reference Bagheri Sheykhangafshe, Rezazadeh Khalkhali, Savabi Niri, Zolfagharnia and Mikelani2024) and binge eating disorder (BED) (Mardi et al., Reference Mardi, Zabihi and Esmailzadeh2025). Positive changes were reported for ED symptoms, quality of life, psychological distress and coping styles; however, EMS and schema modes were not assessed. Group content favoured cognitive and behavioural strategies over experiential content.
Specialist Psychotherapy with Emotion for Anorexia in Kent and Sussex (SPEAKS) is another AN-focused treatment that draws on schema therapy. SPEAKS integrates elements of schema therapy and emotion-focused therapy and has been found acceptable to patients, with promising data on clinical effectiveness (Oldershaw et al., Reference Oldershaw, Basra, Lavender and Startup2023). SPEAKS makes use of a schema mode formulation, but emphasises emotion-focused therapy as the primary change mechanism (Oldershaw et al., Reference Oldershaw, Basra, Lavender and Startup2023).
A schema-informed CBT model for AN
The above studies highlight the relevance of schema concepts to EDs and the emerging evidence for schema therapy with EDs. However, prior approaches either attend to schemas in a largely cognitive way within CBT-ED (Waller et al., Reference Waller, Cordery, Corstorphine, Hinrichsen, Lawson, Mountford and Russell2007a), propose schema therapy principles for the range of ED presentations (e.g. Simpson and Smith, Reference Simpson and Smith2019), or propose a primarily emotion-focused change focus (SPEAKS; Oldershaw et al., Reference Oldershaw, Basra, Lavender and Startup2023). We see four key arguments for considering schema-informed CBT specifically for AN. First, differences in EMS and schema modes have been found across ED diagnoses (Mitchell et al., Reference Mitchell, Huckstepp, Allen, Louis, Anijärv and Hermens2024; Luck et al., Reference Luck, Waller, Meyer, Ussher and Lacey2006), suggesting that schema therapy may require a different focus for AN relative to other diagnoses. Second, treatment outcomes are particularly poor for adult AN relative to other EDs (Søeby et al., Reference Søeby, Gribsholt, Clausen and Richelsen2024), highlighting the need for further improvements in treatment. As above, there is emerging evidence for schema-informed CBT in other presentations of complex need where first-line CBT has not been beneficial. Third, starvation and weight loss have significant effects on physiology and emotion and therefore require attention in AN treatment. CBT techniques are effective for these areas of focus (e.g. Waller et al., Reference Waller, Cordery, Corstorphine, Hinrichsen, Lawson, Mountford and Russell2007a). Fourth, existing AN therapies are already relatively long-term in length (20–40 sessions; National Institute for Health and Care Excellence, 2004), which makes it feasible to consider schema-informed practice as part of existing service timelines. In their guidelines for complex intervention development, the UK Medical Research Council (MRC) suggest that one approach to new intervention development is the extension and adaption of existing treatments (i.e. schema-informed CBT) to new settings or groups (in this case, AN) (Skivington et al., Reference Skivington, Matthews, Simpson, Craig, Baird and Blazeby2021).
Schema informed CBT also holds promise as a neurodiversity-affirming therapy (Spicer et al., Reference Spicer, DeCicco, Clarke, Ambrosius and Yalcin2024; Vuijk et al., Reference Vuijk, Turner, Zimmerman, Walker and Dandachi-FitzGerald2024). Many neurodivergent individuals will experience early unmet emotional needs tied to the experience of trying to navigate a neurotypical world. Schema therapy offers a way of understanding these impacts and associated links with EMS and coping patterns. This is relevant to EDs given the well-documented overlap between autism and AN (Westwood and Tchanturia, Reference Westwood and Tchanturia2017), growing evidence for the over-representation of all EDs in autistic and ADHD individuals (Adams et al., Reference Adams, Mandy, Catmur and Bird2024; Makin et al., Reference Makin, Zesch, Meyer, Mondelli and Tchanturia2025), and poorer treatment outcomes for neurodivergent individuals with EDs compared with their neurotypical counterparts (Adams et al., Reference Adams, Mandy, Catmur and Bird2024). Schema-informed CBT may therefore be one way to improve treatment outcomes for neurodivergent individuals with EDs.
In sum, schema-informed CBT for AN may improve treatment outcomes for this under-served patient group without extending the time frame of existing treatments. To develop a schema-informed CBT model for AN, we created a working group consisting of five highly qualified psychologists/psychological therapists trained in schema therapy and CBT and one lived experience expert with experience of longstanding AN and the benefits of a schema-informed therapeutic approach. The five therapists had each worked for over a decade in the field of EDs and were trained in the existing evidence-based treatments for AN (CBT-ED, MANTRA and SSCM). Further information on training, accreditation and supervision is provided in ‘Therapists and training/supervision procedures’ in the Method section below. The group drew on the original schema therapy components outlined by Young et al. (Reference Young, Klosko and Weishaar2003), schema therapy stages of working with complexity (Arntz and Jacob, Reference Arntz and Jacob2013), recent considerations of schema therapy as applied to EDs (Simpson and Smith, Reference Simpson and Smith2019), an overall schema-informed CBT framework (Moorey et al., Reference Moorey, Byrne and Ruths2020), and core CBT-ED strategies relevant to the nutritional and medical management of AN (Fairburn, Reference Fairburn2008; Waller et al., Reference Waller, Cordery, Corstorphine, Hinrichsen, Lawson, Mountford and Russell2007a). The five psychological therapists developed an initial outline of the proposed schema-informed CBT approach and shared this with the lived experience expert for their thoughts and additional ideas. An assistant psychologist subsequently joined the working group to support the case series evaluation. All co-authors actively contributed to the write-up of the paper. The resulting schema informed CBT model for AN is summarised in Table 1.
Proposed schema-informed CBT treatment model for anorexia nervosa (AN)
SMI-ED, Schema Mode Inventory for Eating Disorders; YSQ, Young Schema Questionnaire.
1 For a review of core schema therapy experiential techniques, see Heath and Startup (Reference Heath and Startup2020). For how to integrate working with the body into time-limited schema informed work, see Briedis and Startup (Reference Briedis and Startup2020).
This model is intended for patients with DSM-5 AN (BMI <18.5) or ‘atypical’ AN who have had prior evidence-based ED treatment and are medically stable for out-patient psychological therapy. The therapy approach may be particularly useful for those with co-occurring psychiatric diagnoses, a history of trauma, and/or chronic unmet emotional needs and complex relational patterns, as identified through clinical assessment or self-report questionnaires. The broader mental health literature indicates that targeting single symptoms where there are broader personality level features leads to poor levels of clinical change (e.g. Goddard et al., Reference Goddard, Wingrove and Moran2015). We propose approximately 9 months of weekly sessions, which would equate to roughly 3 months per therapy stage (see Table 1). In most cases this will equal around 30 weekly sessions, but some patients may require slightly fewer sessions and some slightly more. We expect most people seen with this approach would have between 25 and 40 sessions, with any follow-up period in line with usual service procedures. With the importance of early change in mind we would suggest progression beyond 20 sessions is only indicated where there has been some meaningful change in ED symptoms and/or schemas or schema modes. Future research can help specify reasonable change expectations on suitable measures by this point (Chang et al., Reference Chang, Delgadillo and Waller2021)
As Table 1 highlights, the proposed model places emphasis on a schema therapy mode formulation that incorporates Vulnerable Child, Healthy Adult, maladaptive coping and Critic parts. Example formulations can be found in Arntz and Jacob (Reference Arntz and Jacob2013) and Edwards (Reference Edwards2022). The process of devising a mode formulation with an individual is both normalising and compassionate in that it is a model of the ‘self’ rather than a model of symptoms. It makes sense of ‘coping modes’ as a way to cope at the time of their development (typically early in life) that may now not fit the terrain or needs of adult life. Reliance on these coping modes under-estimates other parts of the self, especially adult parts (the Healthy Adult). Thus, there may need to be a ‘dialling down’ on reliance on some parts of self and a dialling up of others. Therapy involves understanding and modifying the parts of self, including the relationship between these parts, via traditional and experiential CBT strategies (such as schema level continuum work, diary keeping, positive data logs, behavioural experiments, cognitive chair work and imagery work to build up compassion) (Arntz and Jacob, Reference Arntz and Jacob2013). Methods are also used to generate greater ‘emotional heat’ as therapy progresses, to broaden the ‘emotional window of tolerance’, allowing the patient to increase their ability to respond to stressors with adaptive coping. Techniques are also used to explore relationships between parts of the self. Body/somatic methods may also be used to facilitate connection with different parts of the self and to gather a physical ‘signature’ that is elaborated for each mode (Briedis and Startup, Reference Briedis and Startup2020). Eating and weight are considered, but are regarded within a whole-self approach that extends the focus beyond these areas. In more recent applications of schema therapy, internalised critical messages are conceptualised as ‘Critics’, with distinctions between a demanding and punitive Critic. Listening carefully to the nature of the Critic is important and can elucidate various other forms including guilt-inducing, shaming and invalidating Critics. We use ‘The Critic’ as a broad term but clinically it is important to understand the specific content type of the Critic for each individual.
The treatment phases should not be viewed as a list of techniques to be followed but rather as options to draw from based on the stage of treatment and individual’s window of emotional tolerance. We view the change techniques more typically associated with schema therapy – imagery work, chair work and limited reparenting – as being on a continuum from primarily cognitive in nature, right through to ways of working that could be considered ‘emotionally intense’. For example, chair work can be primarily cognitive when we use two chairs to explore pros and cons of AN. Alternatively, we could ‘turn up’ the emotional temperature if using chair work as a springboard to connect back to the early origins of a schema and rescript associated early memories using reparenting informed by unmet need (Heath and Startup, Reference Heath and Startup2020). In a schema-informed CBT approach, we propose that all clinical change tools are used flexibly and with clear psychological targets in mind and tailoring to the individual’s ‘emotional window of tolerance’. Cognitive and behavioural change may precede the deeper emotional change work. In some cases, it may not be appropriate to progress to deep emotional change work, for example where there is not a sufficiently resourced Healthy Adult to moderate and soothe a very fragile child mode, as can be the case for those who have endured early relational trauma.
Similarities and differences between the proposed schema-informed CBT for AN and existing evidence-based AN therapies
Schema-informed CBT for AN differs from existing evidence-based AN therapies in the use of a dynamic ‘whole self’ formulation that is linked to early experiences of unmet emotional needs; the conceptualisation of AN symptoms as part of an array of behaviours which sit within schema coping modes; attention to the Critic in all phases of treatment; greater use of experiential methods like imagery and chair work; the use of the therapeutic relationship as a mechanism of change, including via limited reparenting; and attention to ‘schema chemistry’ within and outside of the therapeutic relationship.
We believe the above features allow for clear differentiation between schema-informed CBT and CBT-ED, MANTRA and SSCM (i.e. the first-line AN therapies). There is greater overlap between schema-informed CBT and SPEAKS, but as noted, SPEAKS places emphasis on emotional processing and emotion-focused therapy as the primary proposed change process. Schema-informed CBT emphasises cognitive and behavioural change processes as primary, with emotional change being a secondary outcome. These different emphases may suit different patients and also different therapists, depending on their core training.
With regard to the therapeutic relationship, many individuals with AN report childhood relational trauma (Elwyn et al., Reference Elwyn, Williams, Smith and Smith2024; Madowitz et al., Reference Madowitz, Matheson and Liang2015) and/or ongoing relational struggles in adulthood which are compounded by loneliness and shame associated with living with an ED (Cummings et al., Reference Cummings, Alexander and Boswell2023; Rance et al., Reference Rance, Clarke and Moller2017; Todisco et al., Reference Todisco, Maragno, Marzotto, Mezzani, Conti, Maggi and Meneguzzo2024). For some of these patients, the core relational qualities of therapist warmth, empathy and positive regard may not be sufficient (Moorey et al., Reference Moorey, Byrne and Ruths2020), particularly due to the complexities of trauma interactions (Olofsson et al., Reference Olofsson, Vrabel, Kopland, Eielsen, Oddli and Brewerton2025). Even very subtly misattuned gestures from a well-intentioned therapist can be experienced as crushing, hurtful or dismissive by a patient with a mistrust/abuse schema and/or abandonment schema. Therefore, the therapist’s relational style needs to be calibrated to the individual with close attention to non-verbal cues that may indicate the individual is hyper- or hypo-aroused (Olofsson et al., Reference Olofsson, Vrabel, Kopland, Eielsen, Oddli and Brewerton2025). Unlike traditional forms of CBT, the therapeutic relationship is specifically emphasised as a mechanism of change in schema therapy. Limited reparenting is part of this emphasis, and can be misunderstood as an infantilising relational stance promoting dependency. This misses the subtleties of the application of this relational style, which involves responding to a patient based on sensitivity and attunement to the mode they are in and the developmental age of that mode. This may include using attuned, empathic and compassionate responding when an individual is in a child mode, in a way that is quite different to the responses needed when an individual is in their Healthy Adult mode or a coping mode. Given the shorter time frame of this schema-informed CBT model, there will be a ‘handing over’ of reparenting from therapist to the individual’s Healthy Adult as therapy progresses.
An attuned therapeutic relationship also allows for effective use of empathic confrontation to challenge EMS and unhelpful coping modes. This technique involves naming an aspect of the patient’s presentation or behaviour that is keeping them stuck, in a way that acknowledges the distress underneath. For example, with stuckness around low weight, a therapist might say something like ‘Maintaining this low weight prevents the deeper emotional work of therapy from being able to take place. Of course this way of coping is there for a reason, it takes away the threat that comes from feeling vulnerable and emotionally exposed. In the past that wasn’t safe, and your eating disorder is how you protected yourself. Now we’re finding ways to have your needs met safely, and exploring how you can feel safe to be vulnerable’. Typically the next step is to then use the foundation of the therapeutic relationship as a safe harbour to work together to find a way out: ‘I would love to be able to help you feel safe enough so that these parts of you can receive the care and respect they deserve, and all sides of you can have a say in what you need. To do this we need to find a way to manage your physical health alongside your emotional needs’. In schema-informed CBT, once trust has been established in the therapeutic relationship, empathic confrontation can nudge motivation in the direction of less reliance on coping modes and greater connection with needs and feelings.
With regards to ‘schema chemistry’, this refers to the relational patterns that flow from interactions between different schemas and schema modes. These patterns may play out within an individual and in their social relationships, but also within therapy and clinical services. For example, in the case of AN, someone may become stuck in a Hopeless Surrender mode in response to underlying schemas such as Dependence/Incompetence, Failure and Self-Sacrifice. When in this mode, the patient may surrender hope for change and believe they are incapable of making changes (reduced self-agency). This Hopeless Surrender mode may elicit frustration from staff, which could in turn exacerbate the underlying schemas around incompetence and failure. Considering schema chemistry within the therapy room can help both patient and therapist step back from ‘stuck’ patterns, but schema chemistry can also be considered in supervision and at a team level, particularly when multiple team members are supporting the patient’s care and/or a patient moves between different parts of the service (e.g. from out-patient to in-patient care). The schema-informed CBT model proposes that considering these relational and system level schema chemistry processes is critical to effective and collaborative clinical care for individuals with AN.
Objectives and case series
We propose that schema-informed CBT for AN may allow for improved treatment outcomes. In order to evaluate the proposed model, we conducted a case series in an adult NHS ED out-patient service. Our research questions were:
-
A. Is schema-informed CBT for AN acceptable and feasible?
-
• As indicated by engagement, treatment completion rates and treatment throughput within the intended number of sessions.
-
-
B. Is there preliminary evidence for the effectiveness of schema informed CBT for AN?
-
As indicated by positive changes in BMI and scores on the Eating Disorder Examination-Questionnaire (EDE-Q), CORE-10 measure of depression/anxiety, Young Schema Questionnaire-Short form 3 (YSQ-3) and Schema Mode Inventory for Eating Disorders-Short form (SMI-ED-SF).
-
Method
Design
We developed the protocol for schema-informed CBT for AN in 2019. Due to the impact of the COVID-19 pandemic, application of the model did not begin until March 2021. We used a case series design and enrolled patients from March 2021 to December 2023. Participants were patients with AN or atypical AN who had completed at least one prior evidence-based treatment for their eating disorder; who were suitable for out-patient psychological therapy; and who seemed appropriate for schema-informed CBT by virtue of additional co-occurring diagnoses and/or early life trauma. Eleven patients met these criteria and were offered schema-informed CBT over the time period in question. All of these patients accepted schema-informed CBT.
Data collection was embedded within routine clinical practice. The case series was approved as a service evaluation project (ID 319) by the South London and Maudsley NHS Foundation Trust (SLaM). As such, research ethics was not required. Participants were informed their routinely collected data would be used for service evaluation purposes and could opt-out of this if they wished.
Participants
Baseline patient characteristics are summarised in Table 2. All patients were White British or White European women whose gender identity matched their sex assigned at birth. Eating disorder diagnoses were: n=6 AN restrictive subtype, n=4 AN binge/purge subtype, and n=1 atypical AN. For the patients with AN binge/purge subtype, three reported regular binge eating at baseline (mean episodes/month=35.7), three reported regular self-induced vomiting (mean episodes/month=39.0), and three reported regular laxative misuse (mean number of days of laxative misuse/month=25.3). Four of the 11 patients reported driven exercise behaviour at baseline (mean number of exercise days/month=20.2).
Baseline characteristics of the 11 patients enrolled in the case series
1 Young Schema Questionnaire (YSQ)-3S subscales scoring in the ‘high’ or ‘very high’ range. As one patient did not complete the YSQ-3S, percentages are out of 10.
As per the inclusion criteria, all patients had prior mental health treatment of some kind. Most (n=8/11) had experienced day or in-patient treatment at least once and most (n=8/11) had at least one formally diagnosed co-morbid condition Patients did not receive any other psychological therapy while receiving schema therapy. Two patients were also under the care of a Community Mental Health Team (CMHT) for care coordination, crisis support and/or medication management. Three patients were seen for physical health monitoring in the ED service separately to their psychological therapy (i.e. by a nurse), where physical health risks were deemed too high to be safely managed by a psychological therapist.
Therapists and training/supervision procedures
Treatment was provided by three of the five therapists who developed the schema-informed CBT protocol. These therapists each had over a decade of experience using CBT-ED and were working as senior clinicians within the specialist NHS ED service. All therapists completed schema therapy training via a standard individual adult International Society of Schema Therapy (ISST) Approved Training Program. One of the three therapists progressed through to ISST advanced accreditation. Each therapist saw 3–4 patients. Therapists received group supervision specifically for this case series by an ISST advanced schema therapist, supervisor and trainer. This supervisor brought over a decade of experience in teaching and training CBT therapists in a schema-informed approach, within National Health Service (NHS) and private sector settings. Consideration was given to schema chemistry within the therapy room, but also within the service, such as when considering transition to more intensive treatment. Session recordings were used within supervision where available, but it was not mandatory for patients to consent to their sessions being recorded, and full tapes were not rated for this case series evaluation.
Measures
Eating Disorder Examination-Questionnaire (EDE-Q; Reference Fairburn and Beglin Fairburn and Beglin, 2008)
The EDE-Q is a self-report measure of cognitive and behavioural eating disorder symptoms over the past 28 days. It generates four subscale scores (Restraint, Eating Concern, Weight Concern, and Shape Concern) and a global score (the mean of the four subscales), and assesses for the presence and frequency of binge eating, self-induced vomiting, laxative misuse and exercise for weight/shape control. The four subscales are not well supported by factor analysis, but the measure otherwise has excellent reliability and validity (Fairburn and Beglin, Reference Fairburn and Beglin2008; Mond et al., Reference Mond, Hay, Rodgers and Owen2006). We used the global score as an overall index of ED psychopathology. Scores can range from 0 to 6, with a score ≥2.8 suggesting clinically concerning ED symptoms (Mond et al., Reference Mond, Hay, Rodgers and Owen2006).
Clinical Outcomes in Routine Evaluation-10 (CORE-10; Reference Barkham, Bewick, Mullin, Gilbody, Connell, Cahill, Mellor-Clark, Richards, Unsworth and Evans Barkham et al., 2013)
The CORE-10 assesses for symptoms of anxiety and depression over the past 2 weeks. It generates a total score ranging from 0 to 4, where higher scores indicate greater distress. Scores ≥1.0 suggest clinically significant distress. The measure was developed for use in UK primary mental health care settings and is now widely used across UK mental health services. It shows good internal consistency and validity (Barkham et al., Reference Barkham, Bewick, Mullin, Gilbody, Connell, Cahill, Mellor-Clark, Richards, Unsworth and Evans2013).
Young Schema Questionnaire-Short form version 3 (YSQ-S3; Reference Young and Brown Young and Brown, 2005)
The YSQ-S3 assesses 18 early maladaptive schemas via 90 self-report items. It is a shorter version of the 232-item long form YSQ-L3. It is considered valid and reliable in clinical and research settings (Bach et al., Reference Bach, Simonsen, Christoffersen and Kriston2015). We used the recommended scoring system to categorise subscale scores as ‘low’, ‘medium’, ‘high’ or ‘very high’ (Young and Brown, Reference Young and Brown2005). The 18 different subscales have different score ranges (e.g. ranging from 0 to 54 for Emotional Deprivation but 0 to 102 for Abandonment, Mistrust/Abuse and Self-Sacrifice).
Schema Mode Inventory for Eating Disorders-Short form (SMI-ED-SF)
The 124-item Schema Mode Inventory (SMI; Young et al., Reference Young, Arntz, Atkinson, Lobbestael, Weishaar and van Vreeswijk2007) was developed to assess schema coping modes in individuals with borderline and anti-social personality disorders. The Schema Mode Inventory for Eating Disorders (SMI-ED; Simpson et al., Reference Simpson, Pietrabissa, Rossi, Seychell, Manzoni, Munro, Nesci and Castelnuovo2018) adapted the SMI for individuals with EDs. Like the original SMI, it assesses for Healthy Adult, child and critic modes. Items for these subscales are broadly similar across the SMI and SMI-ED, as are those for the Compliant Surrenderer, Detached Protector, Detached Self-Soother, Self-Aggrandiser and Bully/Attack coping modes. The SMI-ED then differs from the SMI in assessing for Eating Disorder Overcontroller (e.g. ‘Controlling my eating and/or exercise makes me feel powerful’) and Helpless Surrenderer (e.g. ‘It’s too hard to make changes to my behaviour’) coping modes.
The SMI-ED contains 190 items and was first evaluated in a community sample (n=573) with elevated rates of ED symptoms (n=201 reported an ED diagnosis). Confirmatory factor analysis provided support for the proposed subscales and all subscales showed good internal consistency (α>0.80) and convergent validity with the EDE-Q. The SMI-ED-SF was created to provide a shortened measure with 64 items. Evaluation in an Italian sample (n=649) confirmed the factor structure and convergent validity of the SMI-ED-SF but found somewhat lower internal consistencies for the subscales (α>0.64) (Pietrabissa et al., Reference Pietrabissa, Rossi, Simpson, Tagliagambe, Bertuzzi, Volpi, Fava, Manzoni, Gravina and Castelnuovo2020). We nonetheless used the short version to minimise patient burden. We report mean scores which can range from 0 to 5 for each subscale.
Qualitative feedback
At post-treatment, patients were invited to provide qualitative feedback on their therapy via a treatment evaluation form focused on schema-informed CBT. This form is provided in the Supplementary material. Three questions asked patients to rate their treatment experiences on a 1–5 scale (5=high): how safe and comfortable they felt talking in therapy, how useful they found the sessions, and if they used new skills/strategies resulting from therapy. A further question asked patients to rate how they found the length of treatment, from 1=too short through to 3=just right and 5=too long. There were then five open-ended questions inviting feedback on what was most and least liked, particularly helpful content, ideas for change, and any other feedback.
Procedures
Given the time frame of the study and evolving COVID-19 situation, therapy was provided via a mix of online (video) and in-person sessions. Two patients were seen entirely online, four patients were seen entirely in person, and the remaining five were seen primarily in person but with some online sessions.
Baseline patient characteristics (age, BMI, diagnoses) were reported by the treating clinician, taken from the initial assessment with the service. BMI was collected at assessment and at pre-treatment. Clinicians also reported the number of treatment sessions attended, whether treatment was completed or not, and BMI at post-treatment. Clinicians reported on serious adverse events in the form of any suicide attempts or deaths, but accurate logs were not available for incidents such as deliberate self-harm or the need for medical intervention to manage, e.g. low potassium.
Patients were asked to complete the EDE-Q, CORE-10, YSQ-3 and SMI-ED-SF at pre-treatment and post-treatment. The EDE-Q and CORE-10 are routine outcome measures for the service and could be completed on paper or online. They were sent to patients to complete at assessment and/or prior to their first therapy session. The YSQ-3 and SMI-ED-SF were introduced as measures of schemas and schema modes, and were given to patients in paper form in one of their early therapy sessions and again in one of their final sessions.
The qualitative feedback form was presented as optional, and patients were informed they could return it to their therapist or to the service reception staff if preferred.
The questionnaires and qualitative feedback form were entered and scored by a higher assistant psychologist not directly involved in treatment provision.
Analyses
As an initial case series, all data are presented descriptively with no tests of statistical significance. We report effect sizes (d) for changes in mean scores from pre- to post-treatment, using the formula ([post-treatment mean – pre-treatment mean]/standard deviation of post-treatment mean). We use Cohen’s suggested interpretation of effect sizes, where 0.1–0.3 is a small effect, 0.3–0.5 a medium effect, and ≥0.5 a large effect (Cohen, Reference Cohen1988).
We also calculated Reliable Change Index scores for each patient for the EDE-Q and CORE-10, using the standard formula ([post-treatment score – pre-treatment score])/standard error of the measure) and where a score >1.96 denotes reliable change (Jacobson et al., Reference Jacobson, Follette and Revenstorf1984). For BMI, we defined ‘reliable change’ as >0.5 BMI units. We did not create Reliable Change Index scores for the YSQ-3 and SMI-ED-SF, in part due to a lack of established norms for the SMI-ED-SF and in part as reliable change would only be expected on subscales which were elevated at baseline.
Results
Data availability
All patients completed the EDE-Q and CORE-10 at baseline and 10/11 patients completed the YSQ-3 and SMI-ED-SF at baseline. As shown in Table 2, common maladaptive schemas (as indicated by scoring in at least the ‘high’ range on the YSQ-3) were Pessimism (n=7), Mistrust/Abuse (n=5), Social Isolation (n=4), Vulnerability to Harm (n=4), and Emotional Inhibition (n=4).
At post-treatment, 10/11 patients completed the EDE-Q and CORE-10 and 7/10 completed the YSQ-3 and SMI-ED-SF. Reasons for non-completion were largely to do with the length and burden of the two schema questionnaires. Five patients chose to complete the qualitative feedback form.
Research question A: Is schema informed CBT for AN acceptable and feasible?
No patients discontinued treatment early. Eight patients (73%) completed treatment as out-patients and were discharged outright from the ED service. One patient completed 26 sessions of schema-informed CBT without improvements in ED symptoms and then accepted re-admission for in-patient treatment for their AN. One patient completed 45 sessions of schema-informed CBT with partial improvements in ED symptoms and accepted intensive outreach treatment for their AN. Finally, one patient completed 22 sessions of schema-informed CBT and then accepted a general mental health admission
The mean number of treatment sessions across all 11 patients was 32 (SD=10.28), with a range of 22–55. When excluding the three patients who moved into more intensive treatment, the number of sessions ranged from 25 to 55 (mean=33.25, SD=9.82).
There were no serious adverse events in the form of suicide attempts or deaths.
Qualitative feedback
Of the five patients who completed the treatment evaluation form, three gave a rating of 4/5 for how safe and comfortable they felt talking in their therapy and two gave a rating of 5/5 to this item. For the usefulness of sessions, one patient rated 3/5, three patients rated 4/5, and one patient rated 5/5. All five patients rated the length of therapy as too short (two patients rating 1/5 and three patients rating 2/5, where a score of 3=just right). Linked to this, two patients provided qualitative comments about the length of treatment feeling too short. For example, ‘I just wish I had more sessions as starting to open up more’. Other than longer therapy, the only other suggestion for change was to have face-to-face sessions rather than virtual. This applied to a patient seen earlier in the case series where online working was more common.
In terms of positive feedback, patients spoke of schema-informed CBT providing ‘a safe space to talk about behaviours and support finding practical ways to avoid standard coping responses’ and ‘a place of safety to be vulnerable’. There was broad positive feedback, for example, ‘really liking this type of therapy’. For the most helpful components of therapy, one patient reflected that ‘I found discovering my different coping modes and what can trigger different ones to come into play really interesting’ and others cited ‘identifying schema modes and behaviour patterns’ and being ‘well-supported’.
Research question B: Is there preliminary evidence for the effectiveness of schema-informed CBT for AN?
Changes in eating disorder symptoms
For BMI, 6/11 patients showed improvements from pre- to post-treatment (BMI increase of >0.5 units), 4/11 patients had stable BMI over treatment (BMI within ±0.5 units), and 1/11 patient had a deterioration in BMI over treatment. The patient whose BMI deteriorated was the individual who accepted re-admission for further in-patient care. Patient-by-patient BMI changes are shown in Fig. 1. Mean scores at assessment, pre-treatment, post-treatment and for changes across treatment are shown in Table 3. The overall effect size for group BMI change was small in magnitude.
BMI over time by patient (n=11). The dashed red horizontal line shows BMI 18.5, i.e. the threshold for clinically significant underweight.
Means (and standard deviations) for outcome measures at pre- and post-treatment
1 Mean BMI at assessment=16.38 (SD 1.88); two patients had intensive treatment (in-patient/day patient) between assessment and starting out-patient therapy.
For the EDE-Q, one patient did not complete post-treatment measures. For the remainder, 6/10 patients experienced reliable improvements in Global EDE-Q scores over treatment and 4/10 patients had stable scores (n=3 slight improvements which did not meet Reliable Change Index criteria, n=1 no change). Four patients had scores in the non-clinical range at the end of treatment, although only two patients experienced a reduction in scores from the clinical to non-clinical range (see Fig. 2). Mean scores at pre- and post-treatment are shown in Table 3. The overall effect size for group EDE-Q score change was large.
Global EDE-Q scores over time by patient (n=10). The dashed red horizontal line represents a score of 2.8, the threshold for clinically significant symptoms.
Of the three patients reporting binge eating at baseline, one had stopped by post-treatment and two continued to report binge eating but with reduced frequencies.
Of the three patients reporting self-induced vomiting at baseline, one had stopped by post-treatment; one continued to report self-induced vomiting at post-treatment but at a reduced frequency (15 episodes per month to 4); and one continued to report self-induced vomiting at post-treatment with an increased frequency (18 episodes per month to 28).
Of the three patients reporting laxative misuse at baseline, two had stopped by post-treatment and one continued to report laxative use at the same frequency.
Finally, of the four patients reporting driven exercise at baseline, three had stopped by post-treatment while one patient continued to report exercise at a similar level. The mean number of episodes for each of these behaviours is summarised in Table 3.
Changes in depression/anxiety
Again, one patient did not complete the CORE-10 questionnaire at post-treatment. For the remaining patients, 4/10 experienced reliable improvements in depression/anxiety over treatment, 5/10 experienced broadly stable depression/anxiety, and one patient experienced a reliable increase in depression/anxiety (this being the individual who accepted re-admission for in-patient care). Two patients experienced reductions to the non-clinical range (see Fig. 3). The overall effect size for group CORE-10 change was large (see Table 3).
CORE-10 anxiety and depression scores over time by patient (n=10). The dashed red horizontal line represents a score of 1.0, the threshold for clinically significant symptoms.
Changes in schemas
There was considerable heterogeneity in baseline YSQ-3 scores and in the changes observed over time. The most consistent patterns were decreasing scores on the Pessimism schema (seen to at least some degree for 6/7 patients who provided pre- and post-treatment YSQ-3 data) and the Vulnerability to Harm schema (seen for 6/7 patients). Overall effect sizes for these reductions were medium in size. Medium effect sizes were also seen for Social Isolation, Incompetence/Dependence, Enmeshment and Insufficient Self-Control (see Table 3). Large effect sizes were seen for reductions on the Abandonment, Failure and Subjugation subscales. These stemmed from a subset of the sample (n=2–3) showing very large decreases on these schemas, while other patients had broadly stable scores.
Changes in schema modes
Again, there was considerable heterogeneity in baseline SMI-ED-SF scores and in the changes observed over time. The most consistent patterns were increasing scores on the Healthy Adult subscale (seen for 6/7 patients who provided pre- and post-treatment SMI-ED-SF data) and decreasing scores on the Detached Protector subscale (seen for 7/7 patients). The overall group effects for these changes were large in magnitude (see Table 3). Medium effect sizes were seen for increases on the Angry Child subscale and decreases on the Vulnerable Child, Punitive Critic, Detached Self-Soother, Bully/Attack, Helpless Surrenderer and ED Overcontroller subscales (see Table 3).
Discussion
This paper presents a schema-informed CBT model for AN and an associated case series evaluation. The work builds on prior applications of schema-informed CBT for other presenting difficulties (e.g. Masley et al., Reference Masley, Gillanders, Simpson and Taylor2012; Peeters et al., Reference Peeters, van Passel and Krans2021) and of broader schema therapy for EDs (e.g. Joshua et al., Reference Joshua, Lewis, Kelty and Boer2023).
For the case series, we were interested in acceptability, feasibility, and preliminary effectiveness. Excellent support was provided for the acceptability and feasibility of the model. Specifically, no patients discontinued treatment early (despite usual drop-out from AN psychological therapy being 50%; Solmi et al., Reference Solmi, Wade, Byrne, Del Giovane, Fairburn, Ostinelli, De Crescenzo, Johnson, Schmidt, Treasure, Favaro, Zipfel and Cipriani2021) and we were able to offer the intended number of sessions (mean=32). One patient was seen for 55 sessions, longer than the proposed range of 25–40, but the other 10 patients had between 22 and 41 sessions. Additionally, 8/11 patients completed treatment as out-patients and were discharged from ED care, despite requiring long-term and/or multiple rounds of treatment previously. Three patients accepted more intensive treatment, and their ability to accept and plan for this treatment was seen as very different from prior admissions. All three of the patients had required in-patient admissions previously, including with use of the Mental Health Act, and their ability to consider the need for intensive intervention, and approach this collaboratively, may be seen as a positive Healthy Adult outcome.
In terms of preliminary effectiveness, our results show improvements in ED psychopathology, BMI, and depression/anxiety that are comparable to those in the reported literature. We also found group-level reductions on most subscales of the YSQ-S3, with medium effects, although these overall effects were often driven by a relatively small number of patients experiencing larger changes and other patients having stable scores. Exceptions were on the Pessimism and Vulnerability to Harm EMS, where decreases were moderate in magnitude and present for six of the seven patients with available data. On the SMI-ED-SF, again, moderate changes were seen for most subscales but often reflected change in only a subset of patients. For the Healthy Adult and Detached Protector subscales, changes were large in magnitude and seen more consistently. The increased Healthy Adult scores suggest increased ‘meta-awareness’ of different parts of the self along with an ability to respond to these parts in a considered, conscious way, with an overall increase in being able to tolerate and respond to emotions in an adaptive way. Decreased Detached Protector scores suggest reduced reliance on shutting off or ‘numbing’ emotions. Changes on Child modes showed a moderate decrease in Vulnerable Child scores, which is a key goal of schema therapy, but also a moderate increase in Angry Child scores. This increase may reflect greater awareness of emotions, rights and unmet emotional needs. Ultimately, the goal of schema therapy is to enable the Healthy Adult to respond to perceived injustice or unmet needs but accessing Angry Child emotion may be an important step towards this being possible. This possibility warrants further attention, particularly as prior applications of schema therapy in EDs have not always assessed schema modes.
Changes on the SMI-ED-SF ED Overcontroller were less consistent in this sample and, indeed, scores on this subscale were less elevated at baseline than scores for other coping modes. The SMI-ED-SF and associated ED Overcontroller subscale are relatively new, and evaluation in different samples is relatively limited. All of the patients seen in this case series had strong overcontroller coping modes conceptualised as part of their schema formulations, with overcontrolled behaviours including ED symptoms (e.g. dietary rules) as well as other features (e.g. work-related perfectionism, obsessive-compulsive routines, ‘over-thinking’). For this sample, the SMI-ED-SF did not seem to capture these overcontrolled coping examples fully. It is possible that some of the items on the ED Overcontroller (e.g. ‘Controlling my eating gives me a physical and mental “high”’ and ‘Controlling my eating makes me feel in control of everything’) are less well suited to patients with long-term AN, who may have adapted to controlled eating being a ‘default’ way of coping with life rather than something that is seen to offer immediate benefits. It would be helpful for future studies to continue evaluating the SMI-ED and SMI-ED-SF to see how these perform in different ED groups.
The schema-informed CBT model for AN was premised, in part, on prior research showing differences in EMS and schema modes across different ED diagnostic groups. In this small sample, we still observed considerable heterogeneity in both EMS and schema modes. This is perhaps unsurprising as prior research has shown that individuals with complex presentations often score highly on an array of schemas/schema coping modes, and one of the advantages of schema therapy is that it can conceptualise these via the whole-self formulation (at the level of modes as well as schemas). This raises the question of whether schema-informed CBT is best kept AN-specific or evolves into a transdiagnostic approach, as has been used in prior evaluations of group schema therapy for EDs and applies in CBT-ED. The high diagnostic cross-over in EDs (e.g. from AN to BN or BN to purging disorder) also supports transdiagnostic approaches (Stice et al., Reference Stice, Desjardins, Shaw, Siegel, Gee and Rohde2025). Diagnostic delineation may therefore be less important than formulating accurately the building blocks of the self (via the concepts of unmet need, schemas and modes) and then intervening with this formulation in mind. This fits with the possibility that change anchored by a whole self-formulation may be more robust and holistic than approaches that target a single or discrete number of coping behaviours, which may then be substituted by another unhelpful way of coping (as can be seen when restriction shifts to deliberate self-harm, for example, or binge eating and purging to substance misuse) (Arntz and Jacob, Reference Arntz and Jacob2013).
In our sample, the most frequently endorsed EMS were Pessimism, Mistrust/Abuse, Emotional Inhibition, Social Isolation, and Vulnerability to Harm. This pattern of results differs somewhat from the systematic reviews of EMS by Bär et al. (Reference Bär, Bär, Rijkeboer and Lobbestael2023) and Maher et al. (Reference Maher, Cason, Huckstepp, Stallman, Kannis-Dymand, Millear and Allen2022), which found Unrelenting Standards, Self-Sacrifice and Failure to be most associated with AN. Our sample included individuals who had not benefited from prior first-line ED treatment and who presented with additional co-morbid diagnoses and/or early life trauma. This may have contributed to the different pattern of dominant EMS in our participants. Further research on EMS (and schema modes) in different ED groups seems warranted.
Another learning point from this case series is that treatment length may need to be longer than we have allowed. Patients were very positive about schema-informed CBT but consistently rated the length of treatment as too short. It might be more realistic to think of this type of work as taking 12–18 months, more in line with schema-informed approaches for ‘personality disorder’ (which we view as a presentation of complex emotional and relational need). This could be achieved via 25–40 weekly sessions followed by a tapered follow-up period, and/or by longer weekly sessions in some cases. If this approach can support long-term change and reduce repeated episodes of future treatment, there would be benefits individually, clinically and at a service/cost level. At the same time, given the positive effects sound within this case series, it will be important to evaluate if longer courses of treatment do produce better outcomes and/or result in greater satisfaction with treatment length. In a recent qualitative study, Miller and colleagues (Reference Miller, Hotte-Meunier, Bain, Macdonald, Steiger and Racine2025) found a theme of ‘treatment is never enough’ amongst women diagnosed with EDs and BPD. Participants in this study acknowledged a desire for longer and/or more intensive treatment and some expressed that they wanted treatment to be lifelong. Supporting patients to have a positive therapy ending will be an important treatment goal regardless of overall length.
Next steps for research and clinical practice
As a new model, this proposed way of conceptualising and treating AN requires further empirical attention. This could be via further clinical outcome evaluation, consideration of more in depth qualitative feedback alongside quantitative measures, and further assessment of EMS, schema modes and ED symptoms. The latter focus may help with considering the relative merit of an AN-specific model versus a transdiagnostic ED approach. As noted, we have also identified questions around the optimal length of treatment.
It is important that mechanisms of change are considered in new interventions (Skivington et al., Reference Skivington, Matthews, Simpson, Craig, Baird and Blazeby2021).. We were not able to address this in our case series but the heterogeneity of YSQ-3 and SMI-ED-SF scores suggests that mechanisms of change may differ across patients. Linked to this, further research is needed to identify the best measures of change for schema-informed CBT. As a service evaluation project, our outcome measures were limited to those used in routine clinical practice with EDs (EDE-Q and CORE-10) and routine clinical practice for schema therapy (YSQ-3 and SMI-ED-SF). Measures of personality disorder traits, self-efficacy and quality of life may capture additional areas of change and provide insight into mechanisms of change. Improved measurement approaches may also help with identifying core, common elements of practice across schema-informed CBT for diverse presentations and schema therapy for EDs. These ideas are all consistent with recommendations for complex intervention development (Skivington et al., Reference Skivington, Matthews, Simpson, Craig, Baird and Blazeby2021).
We do not want our proposed model to distract from what works well in transdiagnostic group schema therapy protocols for EDs (e.g. Calvert et al., Reference Calvert, Smith, Brockman and Simpson2018; McIntosh et al., Reference McIntosh, Jordan, Carter, Framptom, McKenzie, Latner and Joyce2016). An advantage of schema-informed CBT is that it may suit existing clinicians trained in CBT ways of working. Accreditation in schema therapy can take many years and is resource intensive for clinicians, supervisors and services. A schema-informed approach may be more trainable and economical than a full schema therapy skill set. However, this possibility needs evaluation with attention to therapist competencies and patient outcomes. One particular challenge may be providing sufficient training in experiential methods, which are likely to benefit from direct teaching/training and are less easy to learn via written material or self-led online training resources. Dedicated training modules on advanced techniques (e.g. chair work, imagery, schema chemistry) may be one way to offer in-depth ‘add on’ training. Further research may also allow for guidance on which schema therapy components are most important for positive outcomes, which could allow for more focused training. Finally, developing a pool of appropriate supervisors would also be an important task to ensure appropriate delivery of newly learned methods. Supervisors would need to be skilled in schema therapy, CBT-ED, and the formulation and treatment of EDs including management of associated risks. There are likely to be specific challenges linked to briefer schema therapy, such as building a strong therapeutic relationship and utilising associated strategies (e.g. limited reparenting) within a condensed timeframe. Given these specific specialist requirements, dedicated supervisor training courses may be an important focus for expanding access to schema-informed CBT. Considering the contexts in which schema-informed CBT is effective for EDs will be an important part of ongoing evaluations (Skivington et al., Reference Skivington, Matthews, Simpson, Craig, Baird and Blazeby2021).
In this study, consideration of ‘schema chemistry’ guided clinical care but was also able to support clinicians in supervision and when communicating with the broader team. This was particularly helpful for those patients who accepted more intensive treatment. For example, consideration of schema chemistry allowed staff to spot when they were being caught up in a patient’s overcontrolled coping mode so that they themselves became very detail-focused. Spotting these patterns helped staff to step back and think about the bigger picture of the patient and their care, realigning with the patient’s Healthy Adult part. Thus, this theme may be important for ongoing evaluations of schema-informed CBT for EDs. Schema chemistry may be a key target in supervisor training and there may also be merit in the development of measures to assess schema chemistry in services.
Limitations
We present an initial case series evaluation with a relatively small number of patients, and with limited diversity in our sample. Extension to larger and more diverse samples (e.g. in terms of gender/gender identity, ethnicity, culture, education level) will be important. As above, there is also a need for research on mechanisms of change, contextual factors that aid the effective implementation of schema-informed CBT, and training and supervision pathways.
Schema-informed CBT has been suggested as a neurodiversity-affirming therapy (Spicer et al., Reference Spicer, DeCicco, Clarke, Ambrosius and Yalcin2024; Vuijk et al., Reference Vuijk, Turner, Zimmerman, Walker and Dandachi-FitzGerald2024) but we did not focus specifically on neurodivergence in our work. Only two patients in this case series had diagnosed neurodivergence although three others were waiting for autism and/or ADHD assessment or related to having traits of these conditions. Schema therapy may need adapting for neurodivergent individuals via tailoring of experiential methods (e.g. physical items to represent parts rather than using chair work) and attention to critical sociocultural messages and ‘masking’ (Spicer et al., Reference Spicer, DeCicco, Clarke, Ambrosius and Yalcin2024). There may be scope for collaboration with existing initiatives like the Eating Disorders and Autism Collaborative (EDAC; Duffy et al., Reference Duffy, Gillespie-Smith, Sharpe, Buchan, Nimbley, Maloney and Tchanturia2025), Eating Disorders Neurodiversity Australia (EDNA; Cobbaert et al., Reference Cobbaert, Millichamp, Elwyn, Silverstein, Schweizer, Thomas and Miskovic-Wheatley2024), and Pathway for Eating disorders and Autism developed from Clinical Experience (PEACE; Tchanturia et al., Reference Tchanturia, Smith, Glennon and Burhouse2020).
Conclusions
This paper proposes a new model of schema-informed CBT for adult AN. The model draws on existing evidence-based schema therapy approaches for complex presentations, together with established CBT-E/CBT-ED techniques. It offers hope for improved clinical outcomes in a patient group often regarded as one of the most difficult to help, including those with AN who present with co-occurring mental health diagnoses or difficulties and those who have tried first-line psychological therapies without experiencing sustained benefits. Case series results provide compelling support for the acceptability of this model and initial support for effectiveness in key areas of change. We recognise the need for this proposed way of working to be further evaluated particularly with regard to unearthing ‘mechanisms of change’, so that we can begin to enhance the hope of individuals presenting to services with complex needs in AN.
Key practice points
-
(1) AN complicated by co-occurring difficulties and underpinned by pervasive unmet emotion need may be best formulated using a ‘whole person’ schema mode map.
-
(2) Schema therapy and cognitive behavioural therapy techniques can be combined to treat complex presentation such as AN as long as clear treatment targets are maintained.
-
(3) There is value in considering an individual’s ‘window of tolerance’ when considering whether to work at the more cognitive or experiential end of the treatment technique continuum.
-
(4) There is value in engaging with ‘schema chemistry’ both in terms of supporting the relation aspect of an eating disorder and improving useful communication and reflective practice within supervision and teams.
-
(5) A schema-informed CBT approach for AN may be a useful addition to eating disorder treatment pathways but further research is needed to develop this possibility.
Supplementary material
To view supplementary material for this article, please visit https://doi.org/10.1017/S1754470X25100196
Data availability statement
The data that support the findings of this study are available from the corresponding author, K.A., upon reasonable request.
Acknowledgements
We are grateful to the patients who provided input to this case series.
Author contributions
Karina Allen: Conceptualization (equal), Data curation (equal), Formal analysis (equal), Methodology (equal), Writing - original draft (equal), Writing - review & editing (equal); Yael Brown: Conceptualization (equal), Data curation (equal), Writing - review & editing (equal); Rosiel Elwyn: Conceptualization (equal), Writing - review & editing (equal); Danielle Glennon: Conceptualization (equal), Writing - review & editing (equal); Leah Holland: Conceptualization (equal), Data curation (equal), Writing - review & editing (equal); Jessica Safadi: Data curation (equal), Writing - review & editing (equal); Helen Startup: Conceptualization (equal), Supervision (lead), Writing - original draft (equal), Writing - review & editing (equal).
Financial support
This research received no specific grant from any funding agency, commercial or not-for-profit sectors. The research was carried out at the National Institute for Health and Care Research (NIHR) Maudsley Biomedical Research Centre (BRC). K.A. is supported by the Medical Research Council grant no. MR/X030539/1 and the Medical Research Council/Arts and Humanities Research Council/Economic and Social Research Council Adolescence, Mental Health and the Developing Mind initiative as part of the EDIFY programme (grant no. MR/W002418/1).
Competing interests
Author H.S. has contributed to the development of other psychological therapies for anorexia nervosa, including the Maudsley Model of Anorexia Nervosa Treatment for Adults (MANTRA), Integrated Group Based MANTRA (MANTRA-IG), and Specialist Psychotherapy with Emotion for Anorexia in Kent and Sussex (SPEAKS). She is also Co-Director of the Schema Therapy School, UK. The other authors declare no competing interests.
Ethical standards
The authors have abided by the Ethical Principles of Psychologists and Code of Conduct as set out by the BABCP and BPS.
Comments
No Comments have been published for this article.