Clinical staging and eating disorders

Clinical staging in diagnostic practice seeks to establish where an individual lies along the continuum of a course of illness and aims to distinguish earlier and milder or more malleable clinical presentations from those that accompany illness persistence and progression.1 The most well-known forms of staging models are those related to clinical oncology, where staging determines treatment choices and indicates prognosis.2

In psychiatry, conventional diagnostic systems have not considered trajectories into or out of illness or the severity of illness, instead relying on cross-sectional descriptions.2 3 For example, the Diagnostic and Statistical Manual of Mental Disorders (DSM), Fifth Edition (DSM-5) and the Eleventh Revision of the International Classification of Diseases do not explicitly consider clinical stages or make any attempt to differentiate earlier illness from later (chronic) forms of illness.4 Most of the criteria that underpin disorder categories in these diagnostic tools have been derived primarily from observations of those with later stage illness, which translates poorly to earlier presentations and thus impedes prevention and early intervention efforts.5

Focusing on eating disorders (EDs), there has been little to no consideration of clinical staging in the diagnostic criteria used to assess patients. For example, anorexia nervosa (AN) has historically only been diagnosed when it has progressed to the point where an individual is exceptionally underweight and has therefore probably been experiencing behavioural disturbances for a significant amount of time. Consequently, making early diagnoses has been challenging6 and early intervention difficult. Conventional diagnostic systems for EDs have also demonstrated poor validity by being unable to comprehensively differentiate clinical presentations. For example, DSM-IV classifications of EDs demonstrated that the most common diagnosis was the ‘residual’ diagnosis of ED ‘not otherwise specified’ (now ‘other specified feeding or eating disorder’ (OSFED)), with a weighted prevalence of 60% in adults7 and around 60%–80% in children and adolescents.8 9 These EDs have been wrongly assumed to be mild and unimportant despite having psychopathology and secondary psychosocial impairment comparable to AN and bulimia nervosa (BN).7 The 2013 revision of the DSM now better distinguishes between different ED presentations, for example, by the addition of binge eating disorder (BED) and including specific examples within the OSFED category (ie, ‘atypical AN’, ‘atypical bulimia nervosa’, purging disorder).10

While reliance on unspecified ED categories decreased with the DSM-5,11 12 these diagnostic systems are still argued to represent only a snapshot of an ED, not the course of the disorder over time, and rarely reflect the clinical reality that ED presentations are unstable and heterogenous.13 In addition, while the DSM-5 incorporates ‘severity’ markers for AN, BN and BED, these are limited to ratings of weight status (AN), binge-eating episode frequency (BED) and compensatory behaviour frequency (BN, eg, frequency of self-induced vomiting). A systematic review found that these severity markers may currently only demonstrate clinical utility for BN, whereas there is limited evidence supporting the use of these single markers for AN and BED.14 Attempts to consider an individual’s stage of illness and relevant recommendations for intervention/treatment remain limited. Clinical staging carries the potential to revolutionise the understanding of mental disorders and bring higher validity to diagnostic frameworks by integrating new research on genetic, biochemical and neurobiological biomarkers of mental illness to better delineate and distinguish clinical presentations, and ultimately use these to move towards attempts for personalised medicine.5

Staging in other areas of psychiatry

In psychiatry, staging paradigms have perhaps been most prominently used for psychotic disorders, with key aims to promote early intervention, reduce the duration of untreated psychosis and improve the timing of treatment interventions.15 The staging model for psychosis, proposed by McGorry and colleagues in 2006 and revised in 2017,16 progresses from stage 0 to 4 with each stage indicating the increased severity of the patient’s clinical phenotype.17 Stage 0 considers the perinatal, parental, environmental or social risk factors (eg, urbanicity, parental psychosis, pregnancy complications) that may increase the risk for psychosis. These criteria have inspired targets for preventative interventions, such as improving mental health literacy and family psychoeducation.16 Stages 1a–c indicate a clinical high risk for psychosis and the emergence of positive and negative symptoms and brief psychotic episodes. Stage 2 focusses on early intervention for the onset of full-threshold first episode psychosis, and stages 3–4 focus on relapsing illness and severe and unremitting psychotic disorders, respectively. More intensive interventions are recommended at each consecutive stage (eg, the introduction of clozapine at stage 3c where former treatments have not been successful).

Staging concepts have been applied to unipolar depression to inform the longitudinal development of the disorder and to understand non-response to treatment.2 18 The longitudinal model has five stages and moves from a prodromal phase, characterised by non-specific or subthreshold depressive symptoms through to chronic, unremitting major depressive disorder. The staging model has been applied in research to plan therapeutic interventions for depression. For example, in line with the longitudinal staging model of unipolar depression, clinical investigations have tested the sequential treatment combination of psychotherapy after response to pharmacotherapy for patients in the residual phase (stage 3).18 Evidence suggests that this method of treatment sequencing is associated with a reduced risk of relapse for major depressive disorder.19 Therefore, considering the course of illness via staging concepts has had therapeutic implications beyond cross-sectional classifications of illness. Staging models of treatment refractoriness in depression have also informed entry criteria for clinical trials of novel treatment approaches.18

Similarly, a four-stage model for post-traumatic stress disorder (PTSD) has recently been put forward by Nijdam et al in hopes to facilitate research and treatment interventions suited to individual illness trajectories.20 Based on clinical and neurobiological research, the model considers the chronological progression of PTSD via staging axes ranging from stage 0 (trauma-exposed, at risk) to stage 4 (unremitting illness of increasing chronicity). The model also incorporates the neurobiological markers, information processing systems, stress reactivity and consciousness dimensions that may be relevant extensions to each stage of illness. For example, increased amygdala reactivity may be a possible neurobiological marker for patients in stage 0. Proposed clinical applications are given based on the model in terms of graded treatment recommendations for each stage of the illness and several suggestions for future research are also put forward. For example, for patients experiencing subsyndromal symptoms of distress (stage 1b), short interventions are recommended, such as mindful relaxation.

Rationale for a staging framework for EDs

For EDs, the rationale for a staging framework stems from various sources of evidence. Neuroprogression refers to evidence of different stage-related brain changes that are associated with the clinical progression of a disorder.21 Having an ED is associated with changes to brain morphology22 and functioning23 and it has been suggested that neuroprogression can lead to more entrenched, chronic symptoms over time in EDs, with longer illness being associated with neurobiological and cognitive changes that impact illness progression.24 It is also important to consider that, over time, there are profound impacts on physical health for those with EDs that are not always reversible; AN can lead to impaired bone health and osteopenia or osteoporosis persisting even after weight rehabilitation, and purging behaviour in AN and BN can cause the erosion of dental enamel.25 26 Clinical research has also indicated that a longer duration of illness may be associated with poorer treatment outcomes, suggesting that intervening early in the course of an ED may prevent more chronic illness.27 Research findings are mixed though, and a recent meta-analysis found no association between duration of ED and treatment outcomes.28 However, this review may have been impacted by marked levels of heterogeneity present across analyses, for example, the type and intensity of treatment varied across the included studies. Thus, while some evidence has demonstrated illness progression and complexity that may be relevant to a staging paradigm for EDs, more research is needed, especially as most research has focused on AN rather than other EDs.29

Previous attempts to review the literature surrounding staging for EDs include Cosci and Fava’s2 systematic review, which identified four distinct phases of illness for both AN and BN, ranging from a prodromal phase (stage 1) to acute manifestations (stage 2), a residual phase (stage 3), into chronic illness (stage 4), with the characteristics of each phase differing between the two EDs. This review included studies in adult populations with DSM defined psychiatric disorders, conducted prior to the release of the DSM-5 where significant changes in the classification of EDs took place (eg, the addition of BED and severity specifiers, ie, mild/moderate/severe illness). Treasure et al24 also reviewed the literature and proposed a staging model of illness for AN which contains four stages ranging from incipient/high-risk stages gradually to persistent forms of illness. They concluded that a staging heuristic could be useful for AN where there is some evidence for the progression of illness and utility of stage-matched interventions. However, less is known about other EDs like BN or BED.

There are various reasons to conduct an updated review to consider how clinical staging can inform ED research and care. First, the outcomes of current treatment interventions for EDs vary and appear to be suboptimal. Recovery from an ED is entirely possible. However, data from predominantly adult samples suggest while the majority of those with AN and BN will eventually recover, the road to recovery can be long, with many years impacted by illness.30 Remission rates increase with longer follow-up, particularly for AN, demonstrating that it is not uncommon to experience an ED for a substantial number of years before reaching recovery.30–32 Adolescents are argued to recover more successfully than adults with AN and family-based treatments generally show high effectiveness for adolescent AN.33–35 However, relapse is common; a systematic review revealed that around 31% of those with AN relapse after treatment, with younger patients with AN at the same risk of relapse as older patients.36 Being able to identify and increase understanding of early-stage presentations via staging concepts may improve early effective management for EDs and shift the trend of EDs becoming persistent, to illnesses that do not become so protracted.37 38

As well as the above, stage-matched interventions for EDs could show therapeutic and economic benefits. For example, clinical staging for psychosis has allowed for the implementation of safer, less invasive interventions for milder, more fluid or malleable earlier stages of illness.39 These include early intervention services and programmes, which have been spread and applied across the world with well-documented clinical and economic benefits.40 A meta-analysis revealed that such services are associated with better outcomes than treatment-as-usual for early-phase psychosis, setting a strong case for extending the reach of these programmes.41 Current attempts to align treatment pathways to a staging model of illness for EDs include the First Episode Rapid Early Intervention for Eating Disorders (FREED) model. FREED builds on theories of biological malleability in early-stage illness and prioritises treatment for young people (16–25 years old) with recent (≤3 years) onset EDs, in order to promote full recovery and prevent chronic illness. FREED is an evidence-based early intervention model and pathway that is tailored to the developmental needs of young people, by providing rapid access to patient-centred well-coordinated care, via adapting National Institute for Health and Care Excellence recommended ED treatments. The evidence for stage-matched treatment interventions for EDs is promising; FREED has shown important clinical and economic outcomes for patients primarily with AN, BN and OSFED.42–45 However, research on early intervention initiatives for EDs is still limited compared with other fields.

As well as guiding early intervention efforts and treatment planning for early-stage illness, a staging model could guide research to refine and improve treatment options and outcomes for people with EDs that endure after initial treatment. For example, a recent, very large network analysis (n=6850) attempted to identify central symptoms/psychopathology across EDs, comparing networks across duration of illness.46 While there were no significant differences in networks found across duration of illness, the authors found that there were differences in central symptoms present in shorter compared with longer duration EDs—cognitive symptoms were more central to shorter duration of illness subgroups whereas behavioural symptoms were central to the medium to longer duration of illness subgroups. Accordingly, the authors suggest that targeting cognitive symptoms such as weight-based self-judgement (‘the over-evaluation of weight and shape’) may be more effective for treating early-stage EDs than for those with a longer illness duration.

Since the most recent review of staging within EDs,24 there have been some attempts to apply staging models to other EDs, for example, via reward processing alterations in BEDs.47 There is a need for an updated review to explore evidence across all EDs as well as AN, and to review new and emerging lines of research.

Re-evaluating the literature may help to better define the stages of EDs, for example, in defining what constitutes a prodromal ED and how this differs from ‘at-risk’ groups or from disordered eating in the general population. For example, McClelland et al reviewed the literature on trajectories into EDs and found a broad range of psychiatric symptomatology that may precede and be indicative of a future ED, but more research is needed to improve and establish understanding of ED prodromes.38 A greater understanding could inform prevention and early intervention initiatives to better target relevant symptomatology and prevent the progression of EDs.

The proposed review

As described above, the aims of clinical staging models are to improve understanding of illness progression and variable treatment response, inspire preventative interventions and novel treatment approaches, and ultimately move towards personalised medicine. Experts have suggested that methods to assess EDs should start to consider illness stage as well as developmental stage to facilitate treatment planning that is individualised and evidence-based.29 Therefore, the primary objective of the proposed review is to explore the extent and nature of evidence supporting different putative illness stages across EDs and across different clinical, biological and neurobiological domains, in order to make clinical recommendations and guide future research. A review of the literature will also help inform and refine current attempts to align treatment pathways to specific ED stages of illness. Our review will consider a wide selection of study designs to investigate the full extent of what is known on each proposed illness stage or proposals of stage models.

A preliminary search of PROSPERO, MEDLINE, the Cochrane Database of Systematic Reviews and JBI Evidence Synthesis was conducted and no current or underway systematic reviews or scoping reviews on the specific topic were found. One protocol for a systematic review of clinical staging across psychiatry, including EDs, was identified via PROSPERO (CRD42021291703). However, the aims of this review are to synthesise presentations of complete or partial staging models across multiple psychiatric disorders. Our scoping review will differ in that its primary focus is on EDs. A recent, unpublished systematic scoping review on staging models in EDs summarised 11 studies explicitly proposing or testing complete or partial staging models.48 The aim of our review is wider and exploratory, including new fields of study and a broader range of evidence. For example, we will include qualitative and quantitative evidence for illness stages across different clinical and neurobiological domains. For instance, if evidence is available that addresses patient, parent and carer perspectives of progression of illness, these will be included.

Review objectives and questions

1. What is the scope, range and nature of research investigating illness staging paradigms for EDs?

2. What evidence is available for specific illness stages, including high risk/prodromal stage, early-stage and late-stage illness across different EDs?

   1. Which different clinical, biological and neurobiological domains of assessment/functioning does this research span across?

3. What evidence is there which demonstrates the implications of different illness stages for treatment and outcomes across the EDs?