Background

Anorexia Nervosa (AN) is a debilitating illness. While there have been many advancements to treatment protocols and outcomes for people with eating disorders, the field acknowledges there remains considerable room for improvement [1]. Positively, manualised treatments such as Family Based Treatment (FBT) [2, 3] and Enhanced Cognitive Behaviour Therapy (CBT-E) [4, 5] have enabled a wider breadth of outcome research and clinical training to be implemented, and the development of fundamental principles for clinical practice and training standards both nationally [6,7,8] and worldwide [9, 10]. In this paper, we explore the evidence for training strict adherence to treatment manuals, considering the benefits, potential reasons for adaptations and limitations. We then outline evidence behind other therapeutic factors which have been demonstrated to impact outcome regardless of which manual is implemented, such as readiness to change and therapeutic alliance. We conclude with implications and considerations for current research, clinical guidelines, training and supervision.

Manualised treatments

Key manualised treatments have been developed for the psychological treatment of eating disorders. For children and adolescents, the National Institute for Health and Care Excellence (NICE) guidelines [9] recommend family therapy, of which manualised versions include FBT [3] and Family Therapy for Anorexia Nervosa (FT-AN) [11]. If family therapy is not appropriate, individual options include CBT-E adapted for adolescents [12] or Adolescent-Focused Therapy (AFT) [13].

For adults with AN, the NICE guidelines suggest Cognitive Behaviour Therapy (CBT) [4, 5], Maudsley Anorexia Nervosa Treatment for Adults (MANTRA) [14] or Specialist Supportive Clinical Management (SSCM) [15], all of which have accompanying manuals. While not listed as a first line option, Focal Psychodynamic Therapy [16, 17] is also suggested as an appropriate alternative if the above options are deemed unsuitable or ineffective. There are also other versions of psychotherapy, such as Interpersonal Psychotherapy (IPT), which achieve equivalent outcomes [18]. While there is certainly a sound evidence base from which to continue to research treatment, training and supervision, the quality of this evidence base remains a debated topic [19, 20].

There are clear benefits to the use of treatment manuals within the eating disorders field and within psychology more generally. One key benefit is they promote the ability to engage in more rigorous research, such as randomised controlled trials (RCTs) which, alongside meta-analyses, are considered the highest forms of evidence [21]. Furthermore, manuals provide clear guidelines for training therapists or those inexperienced within the field, and their accessibility enables broader dissemination beyond in-person or on the job training. Similarly to other manuals like the Diagnostic and Statistical Manual of Mental Disorders (DSM) [22], treatment manuals also help professionals to adopt shared language. This is important to aid communication and a consistent approach within and between treatment teams, which is particularly important given the need for multidisciplinary care for those with AN [23].

Adherence and drift in treatment manuals

Therapist adherence is defined as the accuracy with which the specified protocols of a manualised treatment are implemented in the clinical setting [24]. It is argued to be a key component in training and providing evidence-based treatment, and is a large focus in research, training and supervision [25]. However, it is also recognised that therapists commonly deviate substantially from empirically supported protocols in psychotherapy both generally, and within the eating disorders field specifically [26]. Such deviations from the manuals are labelled as non-adherence or therapist drift, which is defined as where therapists either actively decide to ignore or passively avoid key elements of treatment manuals [27,28,29,30,31]. Non-adherence and therapist drift often have negative connotations [28,29,30]. Non-adherence is cited as a reason for ‘failed Family Based Treatment’ [29]. An implication of this ‘failure’ is that non-adherent therapists feel responsible for poor outcome [32] or families feel responsible for being treatment failures [33]. An alternative hypothesis is perhaps that strict application of a treatment manual may lead to treatment failures. Rather than lying with clinicians or families, the problem may lie with how manuals are taught, implemented and supervised.

It has been repeatedly demonstrated that rates of the adoption of evidence-based manuals by clinicians in the eating disorders field are low. For those practitioners that identify as using manuals, adherence to the manual is also low [34,35,36,37,38]. One study of 40 therapists delivering FBT to young people with AN found not one therapist delivered a therapy that was consistently adherent to the FBT manual [34]. Similarly, another study found that around one third of FBT practitioners deviated substantially from manual adherent treatment [39]. Research has shown that CBT for adults with eating disorders is also commonly delivered in ways that deviate substantially from empirically-supported protocols, with fewer than half of self-defined CBT clinicians using core CBT techniques when delivering treatment [27]. These studies do not discuss if non-adherent practitioners have better or worse outcomes than adherent practitioners.

Why do therapists deviate from manuals?

Potential reasons cited for non-adherence and therapist drift include comorbidities in complex cases, when the patient is perceived as treatment resistant, a lack of training and supervision in the protocol, and therapist preference, anxiety, mood or personality [26, 38]. These reasons are often framed as the therapist prioritising their own needs or preferences over those of their patient. However, a difficulty with maintaining strict adherence to manuals is that some research indicates that the inclusion of key aspects of treatment manuals does not necessarily impact outcome, such as including the family meal in session two [40, 41] or siblings attending FBT [42, 43]. One implication of this is that adaption of manuals, rather than strict adherence, need to be considered in our training and supervision of the use of manuals in clinical practice. Research into adaptations of manualised family therapy treatments has demonstrated promising results. For example, Parent-Focused Treatment (PFT) [42], short verses longer form of FBT [44], Multi-family therapy (MFT) [45, 46], Temperament-Based Therapy with Supports (TBT-S) [47], adapted family admissions [48], brief intensive programmes [49] and carer skills training programs [50]. Given full recovery rates at end of FBT (as measured by > 95% BMI and within one standard deviation of average Eating Disorder Examination (EDE) Global scores) are between 22 and 49% [42, 44, 51], with dropout rates of up to 20% [42, 52], the need to continue developing appropriate adaptations to existing manuals is something which the authors of manuals themselves, as well as the field more generally, acknowledge and promote [53, 54].

It is worth noting that most treatment manuals include statements promoting adaptations and acknowledging that this is required to apply the treatment most effectively in an individualised and real-world clinical context. This is important as clinical research needs to have real-world significance in addition to research based significance. While comorbid psychological disorders like anxiety and depression often mean a person may be excluded from a treatment trial, the reality of clinical work is that the majority of patients will present with comorbidities which may, and should, impact treatment [55,56,57]. Furthermore, adaptation of treatment manuals is often crucial to respectfully and sensitively embrace different cultural backgrounds of the patient, families, and therapist. More recent research has also identified the need to adapt treatment to better work for various populations, including young people or their families who identify as neurodiverse, such as those diagnosed with Autistic Spectrum Disorder [58], young people engaging in self-harm [59], or those who experience gender diversity or dysphoria [60]. Adaptation is also crucial in working in a trauma informed manner, where issues such as attachment and interpersonal style, self-concept and capacity for mentalisation are important considerations [61].

Interestingly, therapists with greater clinical experience are more likely to deviate from treatment manuals, which may be explained by the finding that some therapists view manuals as too inflexible to apply appropriately to complex cases [62, 63]. This may also be a result of applying evidence-informed practice, which promotes a flexible process of balancing best clinical evidence with therapist experience or clinical opinion and patient values and preferences [64, 65]. Therapists newer to a specific model cite a number of reasons why they are less adherent, including inadequate consideration of comorbid conditions, difficulties engaging parents or members of the multidisciplinary team and insufficient training and supervision [24, 66].

Does adherence lead to better outcomes?

When manual adherence is low, it makes sense that outcomes differ from those found in clinical trials, which inherently require a high degree of manual adherence [28, 67]. A review of the wider psychotherapy literature by those in the eating disorders field allows consideration of whether strict adherence to any treatment manual is of benefit to our patients. In reality, it remains completely unsubstantiated whether or not high levels of adherence are related to therapeutic outcome [68, 69]. A meta-analysis of 36 treatment integrity outcome studies in adults concluded that neither manual adherence nor competence, defined as how well interventions were delivered, displayed an effect on outcome that was significantly different from zero [70]. A review of literature in child and adolescent psychotherapy found a small relationship between adherence and outcome in a meta-analysis of 35 studies [71]. However, the authors note that “the finding of a statistically significant association with outcome is tempered by the fact that the small effect suggests that adherence only accounts for just under one percent of variance in outcomes” (p.426). Importantly, neither therapeutic alliance nor patient readiness for change were considered as potential moderating variables. There is also some data that therapists being flexible in adherence may lead to improved treatment effects, at least within psychodynamic psychotherapy [72, 73].

There is an increasing amount of research on manual adherence relating to treatment efficacy within eating disorders specifically. Loeb et al. [74] found no significant association between adherence and outcome in either CBT or IPT for Bulimia Nervosa (BN). Another study found a complex relationship between adherence and outcome in CBT which changed over time [75]. Initial adherence was high in the behaviourally focused early and middle sessions, with therapists becoming less adherent in later sessions. While higher early adherence was related to less binging behaviour, this correlation decreased over time. There was also a notable correlation between high early adherence and increased drop out [75]. Interestingly, while it is noted that there was significant variation in adherence between the four therapists in the study, it is not reported if the less adherent therapists had poorer outcomes. One limitation of this study is the small sample size, although it does highlight an interesting area for future investigation. There is also a need to research the relationship between adherence and outcome in treatment for AN.

Considering complexity: moderating factors

Perhaps not surprisingly, there appears to be a more complex relationship between adherence and outcome than typically assumed. Potential moderators of the relationship between adherence and outcome may include type of therapy (multisystemic or individual), patient diagnosis, therapist competence, therapeutic alliance and patient readiness to change [76].

Within the general psychotherapy literature, it appears that rigidly implementing a therapy approach is ineffective at best and potentially deleterious to outcomes at worst [77]. This is supported by research which suggests that where therapeutic alliance is poor, strong manual adherence is associated with a poorer outcome [76]. One study found that manual adherence to treat panic disorder was unhelpful when patients had low readiness for change, but was not related to outcome in patients with higher levels of motivation [78]. In evaluating CBT for adults with depression, Snippe et al. [79] found that therapist adherence was not related to positive outcome, particularly when patients were less engaged in treatment. Barber et al. [80] found that therapist adherence when working with people with substance abuse difficulties was less strongly related to outcome when therapeutic alliance was strong. Weck et al. [81] found that in three RCTs for depression, social phobia and hypochondriasis, alliance at session one, but not session two, predicted therapist adherence in later sessions, but that adherence had no relationship with future alliance. Such studies also demonstrate that therapeutic relationship and adherence are likely interrelated and bidirectional.

The above research highlights the complex interaction between patient readiness to change, manual adherence and outcome. When patients have low readiness to change, strict adherence to an action-based manual is likely to be ineffective [82]. This is especially pertinent to patients with AN, characterised by its ego syntonic nature in which sustained readiness to change is typically low [83]. It follows that an initial focus of treatment may therefore need to be engagement and building therapeutic alliance and increasing a patient’s readiness to change, rather than working strictly to a manual [84]. One of the major strengths behind family therapy for eating disorders is that even though young people may not want to engage in treatment, parents are often appropriately concerned and highly motivated. However, this by no means negates the importance of engaging the young person in the process. Strategies to build rapport and engage young people are being developed, such as using psychoeducation on the biological effects of starvation (for example, delayed gastric emptying) [11, 85].

While the relationship between adherence and outcome remains unclear, therapeutic alliance consistently predicts, or is at least correlated with, treatment outcome [86]. Baier et al. [87] reports that there are now over 300 studies of the alliance-outcome relationship, with stronger alliance consistently associated with positive treatment outcome across a range of psychotherapies. It is proposed that alliance is a non-specific mediator of change and is important in all psychotherapies [88], with the impact of the therapist relationship on patient outcome estimated to be between 5 and 8% [89]. Similarly, another study found that 97% of variance between the outcome of individual therapists was due to their ability to form better therapeutic relationships with patients [90]. Of course, it is also likely that alliance and adherence are interrelated and reinforce one another during therapy [89].

Parallel processes are a well-documented phenomenon in which therapists, supervisors, multidisciplinary teams and researchers can unknowingly begin to emulate and repeat the same patterns present for their patients [91, 92]. For an illness such as AN, which is characterised by cognitive and behavioural rigidity, higher threat sensitivity and intolerance of uncertainty, we must be careful as a field that we do not unknowingly fall into the same rigid processes. An example of this may be a specialist treatment team who refuses to provide treatment to a 15 year old patient with AN seeking help without support of their family because they strictly adhere to the idea that family therapy is the only appropriate option, despite the NICE guidelines offering suitable evidence-based alternatives.

Implications for training and supervision

The positive impact of alliance on patient outcome, seemingly over manual adherence, is demonstrated both in the eating disorder and wider psychotherapy literature. Additionally, the importance of readiness to change on the appropriateness of manual adherence is also substantiated. As such, this paper questions the focus on strict manual adherence in training and supervision in the eating disorders field. While there remain numerous benefits to the development and application of treatment manuals as previously mentioned, the field may be better served by shifting the emphasis of training and supervision away from strict adherence to a more flexible, yet empirically consistent approach. Perhaps one can envisage manuals in the eating disorders field as a series of core interventions or tasks that should be included in an eating disorders treatment. Recent practice and training standards [7] have alluded to this way of thinking about treatment. These guidelines highlight that some of the core principles of eating disorder treatment may include developing a therapeutic relationship, assessing motivational status, the use of “evidence-based interventions” which include weighing the patient, a focus on modifying the eating disordered behaviour, developing a clinical case formulation and developing alternative coping strategies, amongst others. These guidelines are clearly based on the existing manuals and evidence base, but perhaps suggest a core principle approach rather than strict manual adherence.

One potential method to move away from a stirictly manualised approach to clinical training and practice has been described through Process Oriented Therapy (POT) [93]. POT proposes that treatment, training and supervision should focus on core clinical processes, including motivational enhancement, behavioural activation, cognitive flexibility and emotional regulation skills, rather than manualised content. In this view, training and supervision should focus on identifying specific patient needs through individual and systemic case formulation, rather than the application of specific manuals. One core process discussed is psychological flexibility, as opposed to rigidity. This seems especially important when working to increase psychological flexibility in people with AN. Further research, clinical work and training is already exploring the importance of promoting flexibility in over-controlled individuals through therapeutic adaptations like Radically Open Dialectical Behaviour Therapy (RO-DBT) [94]. Interestingly, POT also highlights the role of the therapeutic relationship and suggests this may model, instigate and promote the internalisation of an ability to be within the present moment, accept difficult experiences and engage in valued action [93].

Constantino et al. [76] offer a further suggestion to rethinking our training and supervision process. They describe Context Responsive Psychotherapy Integration (CRPI) as another alternative to strictly adherent manual training. The authors draw on data suggesting that successful management of alliance ruptures predicts treatment outcome [95], and therefore that training and supervision should help a therapist to identify and respond to the dynamic nature of the therapeutic relationship. They suggest that adjusting therapist actions to patients’ presenting difficulties, context and non-diagnostic characteristics, including the interpersonal dynamics present during therapy, represents a more skilful and nuanced approach that is well supported within evidence-informed practice. They also suggest replacing long-form trainings on single treatment manuals for specific diagnostic problems with briefer trainings that focus on the main principles underlying the manuals. Training would also dedicate time to teaching the clinicians important contextual markers, such as when to move away from the manual (for example, patient hesitancy to change or need to further develop clinical formulation) and empirically supported interventions appropriate for that circumstance (for example, use motivational enhancement techniques) [84].

Adoption of a more flexible core principle approach to treatment has implications for how adherence is assessed. We are in no way calling for an abandonment of treatment manuals or randomised controlled trials evaluating them. These trials, which need strict adherence to assist replicability, provide important data on which techniques should be included in a core principle approach (for example, weighing the patient). The document produced by Hurst et al. [7] could be modified into a checklist form, so that clinicians can assess their adherence to the core principles in treatment. It may also be important to assess common factors of therapy, such as the therapist’s ability to be empathic and to form a therapeutic alliance, as well as specific techniques included in the treatment. Of course, the key assessment is to monitor patient outcome on a regular basis [7]. Assessment of treatment acceptability, in either a strict or flexible form, to both therapist and patient may also be important.

Conclusion

This paper recognises the important advances that treatment manuals have provided our field. Manuals promote rigorous research, provide clear training and clinical guidelines, allow broader dissemination and accessibility, and create a common language and platform from which to research necessary adaptations. Taken together, manuals provide a set of core principles important in the evidence-informed treatment of eating disorders, such as monitoring weight and other markers of health, supporting physical recovery as a means to establish psychological recovery, and utilising a patient’s support network. We know that training in the delivery of manualised treatment does make the clinician more adherent to the delivery of that manualised treatment [69]. However, a growing body of literature from both the field of general psychotherapy and within eating disorders specifically indicates that treatment adherence does not consistently predict patient outcomes and may indeed be contraindicated in some cases. We suggest that clinicians be encouraged in their clinical practice, training and supervision to use the treatment manual alongside individual and systemic case formulation, the client’s stage of change and the therapeutic relationship, with a focus on resolving therapeutic breaches. In combination, these factors allow a clinician to truly engage in evidence-informed practice. We would recommend future research is undertaken regarding formulation-driven adaptation of manuals, and training and supervision that goes beyond strict adherence to treatment manuals in the field of eating disorders.