The current clinical approach to feeding and eating disorders aimed to increase personalization of management

Abstract

Feeding and eating disorders (FEDs) are a heterogeneous grouping of disorders at the mind‐body interface, with typical onset from childhood into emerging adulthood. They occur along a spectrum of disordered eating and compensatory weight management behaviors, and from low to high body weight. Psychiatric comorbidities are the norm. In contrast to other major psychiatric disorders, first‐line treatments for FEDs are mainly psychological and/or nutrition‐focused, with medications playing a minor adjunctive role. Patients, carers and clinicians all have identified personalization of treatment as a priority. Yet, for all FEDs, the evidence base supporting this personalization is limited. Importantly, disordered eating and related behaviors can have serious physical consequences and may put the patient's life at risk. In these cases, immediate safety and risk management considerations may at least for a period need to be prioritized over other efforts at personalization of care. This paper systematically reviews several key domains that may be relevant to the characterization of the individual patient with a FED aimed at personalization of management. These domains include symptom profile, clinical subtypes, severity, clinical staging, physical complications and consequences, antecedent and concomitant psychiatric conditions, social functioning and quality of life, neurocognition, social cognition and emotion, dysfunctional cognitive schemata, personality traits, family history, early environmental exposures, recent environmental exposures, stigma, and protective factors. Where possible, validated assessment measures for use in clinical practice are identified. The limitations of the current evidence are pointed out, and possible directions for future research are highlighted. These also include novel and emerging approaches aimed at providing more fine‐grained and sophisticated ways to personalize treatment of FEDs, such as those that utilize neurobiological markers. We additionally outline remote measurement technologies designed to delineate patients’ illness and recovery trajectories and facilitate development of novel intervention approaches.

Feeding and eating disorders (FEDs) are enshrined in both the ICD‐11 ¹ and DSM‐5 ² classifications as a single grouping, following the merger of two previously distinct sections, feeding disorders and eating disorders (EDs). This recognizes the expression of feeding and eating difficulties in a continuum across the lifespan.

The term FEDs encompasses a broad group of disorders with both distinct and shared phenomenological features, etiological factors and treatment responses ³ , ⁴ , ⁵ . They include anorexia nervosa (AN), bulimia nervosa (BN), binge eating disorder (BED), other specified feeding or eating disorder (OSFED), avoidant restrictive food intake disorder (ARFID), rumination (or rumination‐regurgitation) disorder, and pica ⁴ .

These are common and disabling disorders occurring in people of all genders and cultures. Typical onset spans from childhood to early adulthood, a developmentally sensitive time, with around 15% of FEDs developing by age 14, around 50% by age 18, over 80% by age 25, and practically no new onsets occurring after the age of 30 ⁶ .

Core to all FEDs is a disturbance of eating behavior, often accompanied by various pathological compensatory behaviors (e.g., driven exercise, self‐induced vomiting, misuse of medications to lose weight). These behavioral features are motivated by weight, shape and appearance concerns, or other (e.g., health‐related) concerns about food, eating and weight.

Psychiatric comorbidities (e.g., mood and anxiety disorders) are common and contribute to adverse outcomes ³ , ⁴ , ⁵ . Mortality is raised, with AN having the highest mortality of all mental disorders ⁷ . A recent meta‐analysis of years of potential life lost (YPLL) found that EDs have the second highest YPLL (16.64 years) of all psychiatric disorders ⁸ . Moreover, one in every two to three people with BN or BED are obese or will become obese, with potentially serious metabolic complications.

Much of the burden of FEDs has remained hidden for a long time. A recent extension of the Global Burden of Disease (GBD) study demonstrated that including BED and OSFED in the analysis, in addition to AN and BN, resulted in a revised estimate of the disease burden, which more than doubled ⁹ . Burden for EDs – including BED and OSFED – has been estimated at 85.9 disability‐adjusted life years (DALYs) per 100,000 person years ⁹ . This is around 5% of the DALYs per 100,000 person years for the twelve mental disorders considered by the GBD study in 2019 ¹⁰ . These revised figures are still likely to be an underestimate, as ARFID, rumination disorder and pica were not included ¹¹ .

Numerous systematic reviews and meta‐analyses have reported a substantial increase in ED symptoms in the general population ¹² , as well as significant increases in severity, distress, hospitalizations and demand for services across different FED diagnoses ¹³ , ¹⁴ , ¹⁵ , ¹⁶ , during the COVID‐19 pandemic.

In contrast to other severe psychiatric disorders, first‐line treatments for FEDs are predominantly psychological. This means that a degree of personalization is typically built in via individual case formulations ¹⁷ . ED recovery rates are sub‐optimal, with approximately 50% of patients recovering with best available evidence‐based treatments, and 20‐30% developing long‐term treatment‐refractory illnesses ¹⁸ .

Personalization and precision medicine approaches may allow for improved outcomes, as has been found in other areas of mental health ¹⁹ , but evidence related to FEDs remains limited. Notably, in studies of psychological therapies, possible mediators and mechanisms of treatments are not always defined or assessed in the same way across trials.

Patients, carers, clinicians and researchers are united in seeing the offer of personalized treatments and care as an important indicator of the quality of services for FEDs ²⁰ and as a priority focus for future research ²¹ . We use Deisenhofer et al's definition of personalization ²² , describing it as anything that a clinician does to “select, adapt or adjust treatment to the individual with the goal of improving outcomes”. Precision approaches, which are “algorithmic, quantitative and empirically derived” are seen as a distinct part of personalization ²² .

Advances in bio‐technologies (e.g., neuroimaging, multi‐omics) aim to facilitate personalized medicine and precision approaches to psychiatric disorders in general and FEDs specifically ²³ , ²⁴ , ²⁵ , via multimodal assessments of very large data sets and use of advances in computing, artificial intelligence and machine learning for data management and analysis ²⁶ . Alternative approaches to personalization focusing on individually tailored person‐specific models ²⁷ , ²⁸ , ²⁹ are also emerging in FEDs. However, these advances are not mature yet for application in ordinary practice.

The present paper aims to review systematically the key domains that are usually considered when personalizing treatment for mental disorders ³⁰ , ³¹ , ³² , ³³ , describing their current status as far as FEDs are concerned. Gaps in current knowledge are highlighted, and potential directions for the future of personalization are outlined.

We cover FEDs across the entire age spectrum. As appropriate, we mention issues around race, culture, gender and social disadvantages. The relevant evidence base is much larger for AN, BN, BED and OSFED, compared to ARFID, pica and rumination disorder. Therefore, in many sections of this paper, the term EDs has been retained to reflect that the available evidence focuses on these disorders rather than covering the broad group of FEDs.

SYMPTOM PROFILES

FEDs are characterized by cognitive, behavioral and physical symptoms that cut across diagnostic categories and can be conceptualized as occurring on a continuum.

Cognitive symptoms include over‐evaluation of eating, weight and shape as well as their control (basing one's self‐worth primarily or entirely on these domains), fear of weight gain or fatness, and more general body image disturbances. The latter may include body image distortion (e.g., seeing oneself as larger than objective size) and body image concerns (wanting to be of a different shape or size). These symptoms can present in all ED diagnoses and may occur at less extreme levels in individuals without EDs.

In cognitive‐behavioral models, over‐evaluation of eating, weight and shape is seen as the “core psychopathology” of EDs ³⁴ . In later stages of illness, chronic stress from starvation and social isolation may lead to depression, neuroadaptation and neuroprogression. This is described in the cognitive‐interpersonal model of AN ³⁵ , which includes the inter‐ and intra‐personal consequences of isolation and stress that can accumulate with enduring illness ³⁶ .

Behavioral FED symptoms include dietary restraint (efforts to eat less or follow dietary rules), dietary restriction (actual under‐eating), fasting, both objective and subjective binge eating (feeling out of control of one's eating with/without eating an objectively large amount of food), purging (self‐induced vomiting or misuse of laxatives or medications, including insulin in those with type 1 diabetes mellitus), and excessive or compulsive exercise. Physical manifestations of these symptoms can include weight loss (with or without objective underweight), weight fluctuations or gain, and a range of physical health problems associated with under‐nutrition, binge eating and purging.

Cognitive and behavioral symptoms are heterogeneous within and across FED diagnoses. Latent class and trajectory analyses have identified clusters of FED symptoms which partially map onto current diagnostic categories. These clusters include individuals who are underweight and report fear of weight gain; who are underweight and do not report fear of weight gain; who have binge eating without purging (often co‐occurring with overweight/obesity); who have binge eating and purging together; and who have dietary restriction and body image concerns without being underweight ³⁷ , ³⁸ , ³⁹ . Binge eating and purging commonly cluster alongside depressive symptoms, deliberate self‐harm, and/or substance misuse ⁴⁰ .

Symptom clusters may vary by developmental stage, and interconnectivity between symptoms may increase with age and illness duration ⁴¹ . Thus, personalized care may be supported by considering developmental and illness stage alongside careful assessment of the specific cognitive and behavioral FED symptoms that the individual is presenting.

There is some preliminary evidence that targeting treatment at an individual's particular constellation of symptoms is associated with good outcomes ²⁹ , suggesting that this may be a promising model for personalization of management. Work in this area is beginning to use idiographic network analysis ²⁸ , ²⁹ and causal discovery analysis ²⁷ to assess an individual's real time data – e.g., obtained via ecological momentary assessment. However, the clinical utility of these approaches remains to be established, and some authors have suggested that, in order to achieve truly person‐centred clinical care, these precision approaches need to be supplemented with additional “ecosocial” components including developmental, cultural, social and experiential dimensions ²⁶ .

At present, clinicians can be encouraged to carefully assess individual symptom profiles in order to try and personalize available evidence‐based treatments. Indeed, treatments such as cognitive behavioral therapy (CBT‐ED, sometimes also referred to as enhanced CBT, CBT‐E) and the Maudsley model of anorexia nervosa treatment for adults (MANTRA) already recommend a personalized formulation to guide application of the treatment protocol. It is also worth emphasizing that, across different FED diagnoses and treatments, early response to treatment, within 4‐6 sessions, is the most robust predictor of overall treatment outcome ⁴² . Thus, regular monitoring of symptoms may help with designing and tailoring treatment and lead to improved outcomes.

The Eating Disorder Examination (EDE) ⁴³ is a semi‐structured interview assessing cognitive and behavioral FED symptoms over the past 28 days. It generates four subscale scores with a rating from 0 to 6: restraint, eating concern, weight concern, and shape concern. It has been adapted for use in children aged 6 and over (Child EDE). It is available in English, Chinese, Croatian, Dutch, Finnish, German, Hebrew, Italian, Japanese, Korean, Malay, Norwegian, Persian, Portuguese, Spanish and Swedish. A 28‐item self‐report questionnaire (EDE‐Q) ⁴³ , adapted from the EDE and generating the same subscales, has been validated for use in adults and adolescents, and is available in the same languages.

The Eating Pathology Symptoms Inventory ‐ Clinician Rated Version (EPSI‐CRV) ⁴⁴ is a semi‐structured interview assessing cognitive and behavioral FED symptoms over the past three months. It generates eight subscale scores: body dissatisfaction, binge eating, cognitive restraint, excessive exercise, restricting, purging, muscle building, and negative attitudes towards obesity. A 45‐item self‐report version (EPSI) is also available in English and Chinese.

Detection of FEDs in cisgender men, as well as in gay and bisexual men, can be hindered by assessment models being developed in relation to cisgender women. The EPSI is an example of a measure that performs comparably across cisgender men and women, as well as in gay and bisexual men. In fact, it includes subscales focused on excessive exercise, muscle building and negative attitudes towards obesity, as well as the more typical areas of body dissatisfaction, dietary restraint, binge eating, and purging.

The EDE/EDE‐Q and the EPSI/EPSI‐CRV are less well suited to assessing symptoms of ARFID, while the Nine Item Avoidant/Restrictive Food Intake Disorder Screen (NIAS) ⁴⁵ can be used for this purpose. The NIAS has three subscales (picky eating, low appetite, fear of eating) and is able to distinguish between individuals with ARFID, with other FEDs, and without ED symptoms.

CLINICAL SUBTYPES

In both the ICD‐11 and DSM‐5, the FED grouping includes AN, BN, BED, OSFED, ARFID, pica, and rumination disorder. The DSM‐5 identifies five OSFED sub‐categories: atypical AN, sub‐threshold BN and BED, purging disorder (PD), and night eating syndrome. These diagnoses can be made using the already mentioned EDE and EPSI interviews, as well as the Eating Disorder Assessment for DSM‐5 (EDA‐5) ⁴⁶ , a semi‐structured interview which generates DSM‐5 diagnoses but no subscale scores. Movement between FED diagnoses is common, particularly from AN to BN and between these diagnoses and OSFED ¹⁸ . Movement from BED to atypical AN may occur with bariatric surgery and weight loss medications.

AN is historically recognized by its hallmark features of someone with an emaciated body who eats very little but does not think she/he is thin (or thin enough) nor perceives any risk related to her/his physical condition. The diagnosis of AN has been broadened in current classifications by: a) “relaxing” the definition of low weight, b) excluding the requirement of amenorrhoea, and c) expanding the criteria “fear of weight gain/excessive preoccupation with weight and shape”.

A significantly low body weight for height, age, sex, developmental stage or weight history is required by the DSM‐5 and ICD‐11 for a diagnosis of AN. The assessment of weight status by body mass index (BMI) has shown some limitations ⁴⁷ . Nevertheless, according to the ICD‐11 ¹ , a significantly low body weight means a BMI lower than 18.5 kg/m² for adults and a BMI‐for‐age lower than the 5th percentile for children and adolescents. Importantly, children and adolescents above the 5th BMI percentile may be considered underweight if failing to maintain their personal growth trajectory. Rate of weight loss is also relevant, and can substitute the underweight level if it occurs rapidly (e.g., >20% of total body weight within 6 months).

Fear of weight gain and drive for thinness are considered the “core” features of AN that sustain eating and weight‐related behaviors. However, they are not always explicitly stated by the patients. Current classifications allow for indirect manifestations of these cognitive features based on collateral reporting (by family members/carers) or the presence of behaviors such as dieting, calorie counting, or body checking. The above‐mentioned semi‐structured interview or questionnaire measures can help with thoroughly exploring such behaviors.

AN specifiers have been kept in the DSM‐5 and ICD‐11. They include the restricting pattern/type and the binge‐purge pattern/type (in which regular binge eating and/or purging are present alongside restrictive eating). Significant crossover with progression of the disorder (especially from the restricting type to the binge‐purge type or to BN) limits the predictive validity of these specifiers in terms of course and outcome ⁴⁸ , ⁴⁹ .

International treatment guidelines for AN ⁵⁰ , ⁵¹ emphasize the importance of multidisciplinary care (e.g., medical, nutritional, psychological); weight restoration; psychoeducation; and the involvement of family members/carers when appropriate. In adults, several outpatient psychological therapies are recommended as first‐line interventions: CBT‐ED, MANTRA, specialist supportive clinical management (SSCM), and focal psychodynamic therapy (FPT). Outcomes across these different treatments seem to be similar ⁵² and there is limited evidence on how to choose between them.

Each of these treatments is protocolized but also personalized. CBT‐ED and MANTRA are guided by a personalized formulation, whereas SSCM encourages personalized assessment of “key problems” and psychoeducation, and FPT is guided by patient‐centred focal hypotheses regarding the individual's AN. Clinicians need to use their judgement and patient input to make personalization choices, but there has been little research on how these decisions are made and how they relate to treatment outcomes. In contrast, “therapist drift” (i.e., clinicians failing to deliver evidence‐based treatments despite having the necessary tools) is recognized as common in the management of FEDs ⁵³ . Until further data are available to guide precision medicine approaches, personalization within evidence‐based treatment guidelines is an important principle for all FEDs (“flexibility within fidelity”).

For children and young people (<18 years) with AN, AN‐focused family therapy (FT‐AN, sometimes also referred to as family‐based therapy, FBT) is recommended as the first‐line outpatient option. This can be delivered as a single‐family or multi‐family intervention, and with children/young people seen with or separately to their family members/carers. Briefer formats of FBT seem to work well for most, except for single‐parent families or if the patient has obsessive‐compulsive disorder (OCD) symptoms ⁵⁴ .

In families with high expressed emotion – defined as hostility, criticism and emotional overinvolvement – outcomes are likely to be poorer, and separate therapy for parents and patients or adjunctive parental emotion coaching may be helpful ⁵⁵ , ⁵⁶ . Expressed emotion can be measured in clinical contexts using the Levels of Expressed Emotion Scale ⁵⁷ .

If FT‐AN is not acceptable or effective, alternative outpatient therapies include CBT‐ED or adolescent‐focused psychotherapy for AN. Both are guided by a personalized formulation and encourage family involvement calibrated to the individual and their family.

While outpatient psychological therapy is recommended as the first‐line treatment approach for AN, some individuals will require more intensive treatment. Decisions about when to step up to day‐service, inpatient or residential care should be guided by medical and psychiatric risk, and whether outpatient therapy is effective over the first 4‐6 weeks of treatment. Thus, again, regular symptom monitoring is an important component of personalizing care. There is limited evidence to support psychotropic medications in the treatment of AN, although atypical antipsychotics such as olanzapine are sometimes used to facilitate weight gain.

In BN, the characteristic behavioral features are regular binge eating combined with inappropriate compensatory behaviors to prevent weight gain. In the DSM‐5, binge eating is defined by: a) eating an amount of food, within a limited time period (e.g., two hours), that is larger than most people would eat under the same circumstances; and b) a sense of loss of control over eating. This definition of “objective” binge eating can be distinguished from “subjective” binges, which involve loss of control but not an objectively large amount of food.

In the ICD‐11, binge eating is defined as loss of control over eating and eating notably more or differently than usual, and so may include eating normal portions of foods that are usually avoided. This represents a major difference in the definition of BN (and BED) amongst current classifications: objectively large binges are required by the DSM‐5, while either objective or subjective binges are accepted in the ICD‐11. In practice, both types of binges are reported by people with BN and BED, and similar levels of eating and general psychopathology are found across individuals reporting objective and subjective binges ⁵⁸ , ⁵⁹ .

As in AN, BN involves excessive preoccupation with body weight and/or shape and an undue influence of these aspects on self‐evaluation. This is linked to recurrent inappropriate weight‐control methods to (try and) compensate for binge eating. Compensatory behaviors may vary or be combined, but self‐induced vomiting is the most common. Both purging and non‐purging methods can contribute to a BN diagnosis, with non‐purging behaviors including fasting, extreme dietary restriction, and extreme exercise. Unlike the DSM‐5, the ICD‐11 requires “marked distress” related to the binge‐purge cycle as an essential feature of BN.

Frequencies for binge eating and compensatory methods have been reduced to a minimum of once a week over three months in the DSM‐5 (over one month in the ICD‐11), versus twice a week in earlier classifications. The once‐weekly frequency shows clinical validity ⁶⁰ , ⁶¹ , and has reduced the number of individuals given an unspecified/other diagnosis. BN is distinguished from AN‐binge/purge subtype as underweight is not present ¹ . BN occurs across healthy weight, overweight and obesity ranges, although premorbid overweight/obesity and weight suppression have been found to increase risk ⁶² , ⁶³ , ⁶⁴ , ⁶⁵ .

Treatment guidelines recommend guided self‐help as the first‐line treatment for BN in adults. This typically involves brief supportive sessions with a therapist “coach” to facilitate application of CBT self‐help material. The self‐help material may be digital or book‐based, and supportive sessions may be in person, online or via telephone.

If this approach is unacceptable, contraindicated or does not result in symptom reduction in the first 4 weeks, individual or group CBT‐ED should be considered, guided by a personalized formulation. Fluoxetine continues to be recommended by the American Psychiatric Association (APA)’s guidelines ⁵⁰ in combination with CBT‐ED, and may improve outcomes from CBT‐ED.

For children and young people with BN, BN‐focused family therapy (FT‐BN) is recommended as the first‐line outpatient option. If unacceptable, contraindicated or ineffective, CBT‐ED can be offered, and would again be guided by a personalized formulation and family involvement calibrated to the individual and her/his family.

Interpersonal therapy and dialectical behavior therapy have also been used successfully in BN ⁵⁰ , although with less overall evidence for support than CBT‐ED. Intensive treatment is less common for BN than AN, but may be appropriate in cases of high medical or psychiatric risk or lack of response to outpatient therapy.

Ten‐session CBT for EDs, “CBT‐T”, is not yet recommended in ED treatment guidelines, but has been steadily growing in evidence ⁶⁶ . Intended for use with all non‐underweight EDs, it is modelled on CBT‐ED but emphasizes rapid behavioral change and greater use of experiential techniques. It seems to be most effective in those with high motivation to change ⁶⁷ . Unlike CBT‐ED, CBT‐T is not guided by a personalized formulation, but the approach recommends applying the protocol in a way that is tailored to individual needs (e.g., in terms of which treatment sections to include and in what order). Although no direct comparison between CBT‐T and CBT‐ED has as yet been conducted, outcomes in CBT‐T appear to be comparable to those in CBT‐ED, albeit with half the number of sessions, so CBT‐T is increasingly used in clinical services ⁶⁸ .

After decades of research, BED was recognized as a new FED category in the DSM‐5 and ICD‐11. The introduction of BED has enhanced the diagnostic consistency and accuracy of FEDs ⁶⁹ , ⁷⁰ , ⁷¹ , ⁷² and reduced reliance on unspecified ED categories ⁷² , ⁷³ .

As in BN, BED is characterized by recurrent binge eating at least once per week for three months, associated with marked distress or functional impairment, and not resulting from another health condition or use of medication/substance. Unlike BN, binge eating is not regularly accompanied by inappropriate compensatory behaviors. Body weight or shape over‐concern may be present in BED, but are not considered essential for the diagnosis. Nevertheless, together with marked distress, these features might reflect severity of BED ⁷⁴ , ⁷⁵ , ⁷⁶ .

The DSM‐5 also requires ≥3 additional indicators of binge eating, such as: eating much more rapidly than usual, eating until uncomfortably full, eating large amounts when not physically hungry, eating alone due to embarrassment around eating; and feeling disgusted, depressed or very guilty after eating ² . These requirements have resulted in lower BED prevalence when applying the DSM‐5 compared to ICD‐11 criteria ⁷⁷ .

It has been estimated that 65‐70% of people with BED present with high BMI (e.g., ≥30 kg/m²), but BED is seen across the healthy and overweight/obesity weight ranges ⁷⁶ . Whilst not covered within DSM/ICD criteria, clinically it may be appropriate to focus on loss of control over eating rather than binge eating in younger children, as defining objectively large amounts of food in the context of developmental shifts in nutritional needs is difficult, and younger children tend to have less access to foods in private ⁷⁸ .

Treatment recommendations for BED are similar to those for BN: cognitive behavioral guided self‐help, group CBT‐ED, and individual CBT‐ED. Interpersonal therapy and dialectical behavior therapy are alternative options. There is less specific guidance on applying family‐based interventions to BED in children and young people. In adults, antidepressants (e.g., fluoxetine) and stimulants (e.g., lisdexamfetamine) have been used successfully alongside psychological therapies, although stimulants have mostly been used in individuals with BED who are obese ² . Use of weight loss medications by individuals with BED who are obese is being explored, but research into the impact of these drugs on ED symptoms is in its infancy ⁷⁹ .

OSFED is a heterogeneous category that has been studied as a single entity and with consideration of DSM‐5 sub‐categories. Overall, individuals with OSFED are highly similar to those with AN, BN and BED when considering ED and general psychopathology, psychosocial functioning, and response to treatment ⁸⁰ , ⁸¹ , ⁸² .

Within the OSFED category, PD is defined by recurrent purging in the absence of binge eating in individuals who are not underweight ² . PD is similar to BN in terms of risk factors, symptom profiles and response to treatment, but may be associated with higher premorbid BMI and greater gastrointestinal distress (e.g., nausea) after eating ⁸³ .

Atypical AN is defined by all criteria for AN being met except that, despite significant weight loss, weight is within or above the normal range ² . “Significant weight loss” is not operationalized, and it is noteworthy that the ICD‐11 allows the diagnosis of AN to be applied when someone has lost >20% of her/his body weight but is not yet significantly underweight.

Individuals with atypical AN are more likely to be male and non‐White than those with AN, and to have higher premorbid BMI, but show comparable or higher rates of ED psychopathology, comparable rates of general psychopathology, and comparable or slightly lower rates of medical complications ⁸⁴ .

There are not yet discrete treatment recommendations for those with OSFED. Most guidelines suggest that recommendations for the closest specific FED (e.g., AN, BN or BED) are applied. CBT‐ED and CBT‐T have been successfully used with a range of OSFED sub‐categories. Of note, two versions of CBT‐ED are available and may help personalization: a large trial found that, for OSFED patients with greater complexity (i.e., high levels of distress, perfectionism, low self‐esteem, and interpersonal problems), a broad form of CBT‐ED which encompassed these topics had better outcomes than a focused form, targeting mainly ED cognitions and behaviors. For patients without these complexities, the reverse was true ⁸⁵ . Perfectionism can be measured in a clinical context by the Frost Multidimensional Perfectionism Scale ⁸⁶ . Self‐esteem can be measured by the Rosenberg Self‐Esteem Scale ⁸⁷ , and interpersonal problems by the Inventory of Interpersonal Problems ⁸⁸ .

ARFID involves severe restriction or avoidance of foods, with underweight and/or nutrition deficiency, without marked body image concerns (although individuals may express concern about low weight). It commonly develops in childhood and adolescence, and has high comorbidity with autism ⁸⁹ , ⁹⁰ . Under‐eating in ARFID may be linked to sensory concerns (smells, textures or appearance), apparent lack of appetite or interest in food, and/or fear of adverse consequences (e.g., choking, nausea). ARFID patients are often younger than those with AN and BN, with a higher proportion of males and of medical comorbidities ⁸⁹ .

Treatment of ARFID may be CBT‐based (individually or involving the family) or consist of family‐based or parenting interventions. Research in this area is still evolving, and there is no compelling evidence at present for one psychological treatment over another. Dietetic, medical and wider multi‐disciplinary care (e.g., speech and language therapy) may be helpful, and higher‐intensity treatment may be required in cases with very low weight or other medical complications. There are calls for further research on the effective treatment of ARFID, and recommendations for personalized management are not yet available ⁹¹ , ⁹² .

There is insufficient research on rumination disorder and pica to address these categories in detail. Rumination disorder is characterized by intentional and repeated regurgitation (bringing previously swallowed food back up into the mouth), which may be accompanied by rumination (re‐chewing and re‐swallowing food) or the food being spat out. The regurgitation needs to occur at least several times per week over at least several weeks. Pica is defined as recurrent intake of non‐nutritive, non‐food substances for at least one month. Recent population‐based findings suggest over‐representation of pica in children with autism and developmental delay, and reductions in prevalence from age 3 years onwards ⁹³ . Behavioral interventions may be helpful, e.g., by reducing access to the non‐food substance and providing an appropriate substitute, and dietetic input can support nutritional improvements where needed ⁹⁴ .

Previously, FEDs were considered largely as Western culture‐bound syndromes, occurring mainly in women. Increasingly, there is now recognition that they are a growing public health problem in many non‐Western cultures, e.g. in many Asian countries ⁹⁵ . Internationally, many individuals with FEDs never receive treatment. Men, those from minoritized ethnic groups; and lesbian, gay, bisexual, transgender, queer or questioning (LGBTQ+) individuals are particularly under‐represented in clinical services, even though FED prevalence rates are rising in these groups.

There is emerging evidence that racial discrimination may increase risk for BED ⁹⁶ . Gender identity and sexual orientation intersect with race/ethnicity to influence ED symptoms in ways that are only starting to be understood ⁹⁷ . Clinicians should consider culture and ethnicity, subjective experiences of racial discrimination and cultural identity, gender identity, and sexuality as part of their assessment of FEDs.

The prevalence of underweight is greater in Asian than Western countries. For example, 20‐24% of young Japanese women are underweight (i.e., with a BMI <18.5 kg/m²) ⁹⁸ . This may influence detection of AN, and makes BMI a poor marker of FED presence or severity. Culturally tailored severity classifications for AN might be needed for treatment optimization. There are also well‐documented presentations of underweight AN‐like patients in Hong Kong, Japan and India in whom fear of fatness is not present ⁹⁹ , suggesting that culture may exert a pathoplastic effect on FED symptoms. As most research on AN without fear of fatness was conducted prior to the release of the DSM‐5, these patients may over time move to being diagnosed with ARFID.

In Arab, Middle Eastern and African countries there are relatively few studies on FEDs. Prevalence rates are thought to have increased over time ¹⁰⁰ , ¹⁰¹ , but with scarce historical data. Restricted food intake may be attributed to somatic complaints, rather than body image concerns, to a greater degree than in Western samples ¹⁰² , which may change the emphasis of assessment and treatment. Exposure to Western culture and higher socioeconomic status have been linked to ED symptoms in some but not all studies.

In the US, young Black/African American women and Hispanic/Latina women report higher rates of binge eating, dieting and purging than their White and Asian American counterparts ¹⁰³ , ¹⁰⁴ . Black/African American and Hispanic/Latino men also report higher rates of binge eating than White and Asian American men ¹⁰⁴ . For diagnosed FEDs, some studies have found higher rates of BN or BED amongst Black American, Hispanic and Latin American groups, compared to White Americans, but results are inconsistent ¹⁰⁵ , ¹⁰⁶ , ¹⁰⁷ . Similarly, race moderated response to ED treatment in some studies, while in others outcomes appear similar across Black, Hispanic and/or Latin American groups and their White counterparts ⁹⁷ . However, treatment use may be lower amongst Black and Hispanic/Latin American than White American groups ¹⁰⁶ , which complicates interpretation of cultural differences.

There has been little research on FEDs in Indigenous populations, but emerging research from Australia, New Zealand, the US and Canada suggests that risk for disordered eating may be higher among Indigenous groups compared to the White ethnic group ¹⁰⁸ . There are calls for research to specifically explore FEDs in Indigenous people and for culturally sensitive approaches to assessing and understanding FEDs in these populations. This will require collaboration and co‐production with local Indigenous communities.

Despite being under‐represented in clinical services, transgender, non‐binary and gender expansive individuals appear to be at higher risk for EDs than their cisgender counterparts ⁹⁷ , ¹⁰⁹ . This may stem from the additional psychosocial stress experienced by individuals in these groups, together with body image concerns specific to gender and sexual diversity, including gender dysphoria ¹⁰⁹ .

Most studies of FED treatments have been conducted with samples of (presumed cisgender) women, but there is growing attention, particularly in the US, to providing care that is gender and sexuality affirming, weight inclusive, anti‐racist and trauma‐informed.

SEVERITY

The assessment of severity is a valuable tool to guide treatment decisions and inform expected outcomes and prognosis in FEDs. As there are no available biomarkers, a definite conception of FED severity remains a matter of debate. Severity specifiers for FEDs were introduced in the DSM‐5 and ICD‐11.

In the DSM‐5, severity levels are based on BMI in AN; on the frequency of compensatory behaviors in BN; and on the frequency of binge eating in BED. Four severity grades are described for each disorder: for AN, “mild” (≥17 kg/m²), “moderate” (16‐16.99 kg/m²), “severe” (15‐15.99 kg/m²), and “extreme” (<15 kg/m²) BMI; for BN and BED, “mild” (1‐3 episodes/week), “moderate” (4‐7 episodes/week), “severe” (8‐13 episodes/week), and “extreme” (14 or more episodes/week).

The ICD‐11 only describes a severity specifier for AN, with BMI being the driving characteristic. Two ICD‐11 underweight sub‐categories are proposed: AN with significantly low body weight (BMI 14‐18.5 kg/m² for adults, or 0.3rd‐5th percentile BMI‐for‐age in children/adolescents) and AN with dangerously low body weight (BMI <14.0 kg/m² in adults, or <0.3rd percentile BMI‐for‐age in children/adolescents). A third category, “AN in recovery with normal body weight”, applies to individuals who restored weight to normal but still experience core cognitive and behavioral symptoms of the disorder.

The conception and grading of severity specifiers of the DSM‐5 and ICD‐11 can be considered somewhat arbitrary, and research into their validity and utility is only partially supportive ¹¹⁰ . For instance, inpatient treatment was found to be positively associated with DSM‐5 AN severity sub‐categories ⁷⁵ , but psychiatric morbidity, distress and cognitive impairment were not ¹¹⁰ , ¹¹¹ , ¹¹² , ¹¹³ , ¹¹⁴ . In BN, a meta‐analysis of seven studies found ED psychopathology increasing with the frequency of compensatory behaviors ¹¹⁰ , whilst measures of impairment, psychiatric comorbidities, cognitive functioning and distress differed between BN severity categories in some but not all studies ⁷⁵ , ¹¹⁰ , ¹¹⁴ . For BED, a meta‐analysis of five studies found higher ED psychopathology in more severe categories ¹¹¹ , while varied results were reported for other measures, such as general psychopathology, cognitive functioning, and quality of life ¹¹¹ , ¹¹³ , ¹¹⁴ . There is very limited research on potential severity markers in ARFID, pica and rumination disorder.

Alternative severity indices have been proposed to reflect the transdiagnostic features of EDs, rather than the current disorder‐specific markers ⁷⁵ , ¹¹⁵ . There is some evidence for the clinical utility of a severity index based on low or high over‐evaluation of weight and shape or drive for thinness ⁷⁵ , ¹¹⁶ , or duration of the disorder ¹¹⁴ . However, conclusions are limited by methodological differences across studies, and there is little evidence for any severity markers reliably predicting differential response to particular treatment approaches ¹¹⁷ , ¹¹⁸ .

It is recommended that the DSM‐5 and ICD‐11 severity categories be used as general guidance, but levels of severity be considered in relation to overall psychopathological status, physical risk, and functional impairment.

In addition to markers of severity, motivation to change is often considered when assessing FEDs. It is common for motivation to fluctuate over time and to vary across FED symptoms ¹¹⁹ . For example, someone may be motivated to reduce her/his preoccupation with eating, weight and shape or stop binge eating, but not to cease dietary restraint or increase her/his weight. Self‐reported motivation does not always correlate with behavioral change, and motivation to change typically increases with effective ED treatment ¹¹⁹ . For these reasons, low self‐reported motivation to change should not be a reason to withhold or stop treatment. However, exploring motivation may support collaborative treatment planning.

CLINICAL STAGING

In EDs, the rationale for a staging framework stems from various sources of evidence. Much clinical research has indicated that outcomes are best in children and young people, i.e., where illness duration is typically short, suggesting that intervening early in the course of an ED may prevent a more chronic illness ¹⁸ . A very large network analysis of people with EDs (N=6,850) attempted to identify central symptoms, comparing networks across duration of illness ¹²⁰ . Cognitive symptoms were found to be more central in the shorter duration of illness subgroups, whereas behavioral symptoms were more central in the medium‐to‐longer duration of illness subgroups. These findings provide important pointers towards stage‐matched interventions, suggesting that targeting cognitive ED symptoms such as the over‐evaluation of weight and shape may be more effective for treating early‐stage EDs than for those with a longer illness duration.

Development and maintenance of an ED is associated with complex changes to brain morphology ¹²¹  and functioning ²⁵ . As in other psychiatric disorders ¹²² , ¹²³ , the brain changes resulting from neuroprogression – i.e., reduced neurogenesis and neuroplasticity – have been postulated to drive clinical progression of EDs ¹²⁴ , ¹²⁵ . It is also important to consider that, over time, there are profound impacts of EDs on physical health that are not always reversible, such as impaired dental and bone health.

Several staging models for EDs have been developed ¹²⁶ . Most of them focus mainly on AN, except one which focuses on binge‐type EDs ¹²⁷ . Criteria for stage differentiation are mostly based on combinations of cognitive, behavioral and physical features, and illness duration.

Treasure et al ¹²⁴  proposed a four‐stage model for EDs, based on that of McGorry et al ¹²⁸ for psychosis, i.e., ranging from incipient/high‐risk stages to persistent forms of illness. The model also suggested stage‐matched interventions. Aspects of this model were tested in a sample of 187 AN patients treated in outpatient services and followed up for one year ¹²⁹ . Despite similar baseline BMIs, early‐stage AN patients had much better BMI outcomes and improvements in work and social adjustment at 12 months, and were less likely to need day‐ or inpatient treatment over the 12‐month period.

Other authors have focused on testing stage‐appropriate interventions for particular populations, in particular early‐stage and severe enduring illness. For example, a service model for early‐stage illness – targeted at young people (age 16 to 25) with a recent (≤3 years) onset ED – is the First Episode Rapid Early Intervention for Eating Disorders (FREED) model ¹³⁰ . This model offers rapid access to youth‐friendly evidence‐based intervention tailored to illness stage (i.e., emphasizing biological malleability and need for early nutritional change), together with developmentally appropriate adaptations to interventions (focusing on identity development and “adulting”, using parental support as appropriate), to promote full recovery and prevent chronic illness. Compared to usual care, FREED has been shown to shorten duration of untreated illness (time from onset to first evidence‐based treatment), substantially improve clinical outcomes, lead to cost savings ¹³⁰ , ¹³¹ , ¹³² , ¹³³ , and be highly acceptable to patients ¹³⁴ .

With a focus on AN, others have operationalized the severe and enduring illness stage, with a proposed set of empirically testable criteria (illness duration >3 years and at least two unsuccessful evidence‐based treatments appropriately delivered) ¹³⁵ . A Cochrane review ¹³⁶ identified one RCT of psychological interventions (CBT and SSCM) adapted for this chronic population, with no differences in outcomes between them.

In conclusion, staging models for FEDs are promising, but still in their infancy, with few agreed definitions. Current stage characterizations focus mainly on clinical features, but will ultimately be able to truly transform clinical care only if they can be extended to include genetic, neurobiological and other physical markers of illness risk, treatment outcome and prognosis ²⁵ .

PHYSICAL COMPLICATIONS AND CONSEQUENCES

Physical sequelae of FEDs can affect any organ system. They mostly arise either from self‐starvation and low weight (as in AN, ARFID or atypical AN) or from purging (as in binge‐purge AN, BN or PD), or are related to persistent binge eating and overweight/obesity (as found most commonly in BED). The risk of serious physical health consequences, including death, is typically high‐est when self‐starvation and purging occur together. An overview of starvation‐ and purging‐related complications is provided in Table 1.

Table 1.

Physical complications of malnutrition and purging behaviors

	Malnutrition		Purging
	Pathological findings	Symptoms	Pathological findings	Symptoms
Central nervous system	Decrease in the volume of grey and white matter and functional alterations in the brain	Cognitive impairment	Swollen brain cells (e.g., due to low plasma sodium)	Epileptic seizures
Endocrine system and reproductive function	Hypogonadotropic hypogonadism, growth hormone resistance, low levels of IGF‐1; low levels of TSH, T3 and T4; hypercortisolaemia, hypercholesterolaemia, hypotestosteronism in males	Amenorrhoea, hypoglycaemia, growth delay in children and adolescents
Cardiovascular system	Low blood pressure, bradycardia, hypokalaemia and QT prolongation	Asthenia, syncope, cardiac arrhythmia	Hypokalaemia, acid‐base disorders and QT prolongation	Cardiac arrythmia
Teeth and parotid glands	Dental damage	Dental caries	Erosion of lingual dental surface, parotid glands hypertrophy	Irreversible dental caries, parotid glands inflammation
Gastrointestinal (GI) tract	Impaired gastric emptying, low GI motility, altered gut microbiota	Constipation, early fullness	Oesophagus and pharynx mucosal laceration, low colon motility, hyperamylasaemia	Cough and dysphagia, episodes of epistaxis, diarrhea/constipation, nausea, abdominal pain
Urinary tract	Hypokalaemia, hypochloraemia, metabolic alkalosis	Kidney damage	Hypokalaemia, hypochloraemia, hyponatraemia, metabolic alkalosis, hyperaldosteronism	Kidney failure, edema
Respiratory system			Intrathoracic pressure increase	Pneumomediastinum and aspiration pneumonia
Haematological and immune system	Reduced bone marrow cell production, hypoproteinaemia	Anaemia, susceptibility to bacterial infections, compromised immune function, edema
Bone	Osteopenia, osteoporosis	Bone fractures
Muscle	Myopathy	Impaired muscle strength (e.g., leading to inability to get up from a squat)	Decreased muscle blood flow	Muscle cramps
Skin	Dystrophic and dry skin, lanugo hair	Fragile and fulling hair, acrocyanosis	Russell's sign (calluses on the knuckles or back of the hand due to repeated self‐induced vomiting)

IGF‐1 – insulin‐like growth factor 1, TSH – thyroid stimulating hormone, T3 – triiodothyronine, T4 – thyroxine

The mortality from both medical morbidity and suicide is high for all EDs ¹³⁷ , with standardized mortality ratios of 5.86 for AN, 1.93 for BN, and 1.92 for OSFED. One in five of AN deaths result from suicide. Large national register studies confirm these findings ¹³⁸ , with some pointing to a gender difference, i.e., with higher mortality in men than women with AN or BN ¹³⁹ . For BED, the standardized mortality ratio is 1.50 to 1.77 ⁵ .

In a longitudinal hospital register cohort of 5,169 women, AN was associated with death from suicide (hazard ratio, HR=4.90, 95% CI: 1.93‐12.46), pulmonary diseases (HR=3.49, 95% CI: 1.77‐6.89), diabetes mellitus and other endocrine diseases (HR=7.58, 95% CI: 1.89‐30.42), liver and other gastrointestinal diseases (HR=3.27, 95% CI: 1.33‐8.06), and shock and organ failure (HR=3.59, 95% CI: 1.23‐10.49). Among pulmonary causes, AN was most strongly associated with death due to pneumonia (HR=8.19, 95% CI: 2.78‐24.14) ¹⁴⁰ .

The APA ⁵⁰ and the UK National Institute for Health and Care Excellence (NICE) ⁵¹ guidelines recommend that, in addition to a careful mental health and physical history, the initial physical examination of a patient with a possible FED includes assessment of vital signs (i.e., temperature, resting heart rate, blood pressure, orthostatic pulse and blood pressure); height, weight and BMI; and signs of malnutrition and purging. These guidelines also recommend a complete blood count and a comprehensive metabolic panel, including electrolytes, liver enzymes, and renal function tests. An electrocardiogram should be obtained in patients with a restrictive ED, in those with severe purging behavior, and in those taking medications known to prolong QTc ⁵⁰ . Weight measurements will typically be recorded on a session‐by‐session basis during treatment, with other examinations (including height in children) and investigations recommended at regular intervals, depending on medical risk, presence and severity of physical comorbidities, and treatment response. Guidance on the recognition and management of medical emergencies in EDs has been provided by the UK Royal College of Psychiatrists ¹⁴¹ .

Most physical consequences of starvation are reversible with weight regain and normalization of eating. An exception are the effects on bone density. NICE guidelines ⁵¹ recommend that a bone mineral density scan is considered after one year of low weight in children and adolescents and after two years in adults, or earlier if there are bone pain or fractures. Patients with a persistently low weight can suffer an annual loss in bone density of up to 10% ¹⁴² , ¹⁴³ . The best treatment is reaching and maintaining a healthy BMI for age. Oral oestrogens, calcium‐vitamin D3 preparations and bisphosphonates are unlikely to halt or improve the reduction in bone density whilst low weight persists. More recent studies point to the usefulness of transdermal oestrogen application in adolescents and of bisphosphonates in adult patients with persistent AN ¹⁴⁴ . However, bisphosphonates have potentially teratogenic effects that need to be explained to patients, so they can weigh up risks and benefits.

Bone marrow suppression may lead to anaemia and leukopenia, both of which normalize with weight gain ¹⁴⁵ . Liver damage, as reflected in raised transaminases, is highly correlated with low BMI and again quickly normalizes with weight gain ¹⁴⁶ .

In patients with AN and ARFID, a wide range of upper and lower gastrointestinal symptoms are common in the underweight state, with constipation, nausea and abdominal pain the most common in AN ¹⁴⁷ . These often improve during or after nutritional rehabilitation ¹⁴⁸ . The gut microbiome may be involved in the pathophysiology of gastrointestinal symptoms in AN, as it is affected by alterations in energy intake and dietary composition. In some cases, probiotics have shown therapeutic effects ¹⁴⁹ . As research in this area continues to progress, there may be new opportunities for personalizing care, including either direct changes to gut microbes (via probiotics or faecal microbial transplants) and/or changes to the microbial environment (prebiotics, diet) ¹⁵⁰ , ¹⁵¹ .

AN also leads to endocrine changes, such as hypothalamic amenorrhea, low triiodothyronine (T3) and thyroxine (T4), low levels of insulin‐like growth factor 1 (IGF‐1), relative hypercortisolaemia; decreases in leptin, insulin, amylin and incretins; and increases in ghrelin, peptide YY and adiponectin ¹⁵² . Correction of thyroid hormones is not advised, given the potential for misuse of these hormones to aid weight loss. Most of these changes are adaptive and reversible with weight restoration, although a degree of growth stunting may persist and for some patients there is Cushingoid truncal weight gain during re‐feeding ¹⁵² .

Purging can lead to many physical consequences, that vary according to the type and frequency of purging behaviors. The most dangerous and life‐threatening medical complications are cardiac arrythmias and QT prolongation due to electrolyte imbalance, especially hypokalaemia and acid‐base disorders ¹⁵³ . Severe hypokalaemia can also promote renal failure. Vomiting and diuretic abuse are associated with the development of hypokalaemia, hypochloraemia, and metabolic alkalosis; laxative abuse can present with hypokalaemia and hypochloraemia ¹⁵⁴ . Hyponatraemia can occur as a consequence of all purging behaviors ¹⁵⁴ . Assessment and regular monitoring of serum electrolyte disturbances and electrocardiogram evaluation must be considered essential steps in the management of patients with purging behaviors. Potassium depletion not lower than 2.5 mmol/L without symptoms or electrocardiographic changes can be treated with oral potassium supplementation and correction of volume depletion, while intravenous repletion of potassium is required at <2.5 mmol/L levels ¹⁵⁵ .

In addition to the above life‐threatening complications, stimulant laxative abuse can cause reduction of gastrointestinal motility, and patients may present with chronic diarrhoea, constipation, nausea, or abdominal pain ¹⁵⁶ . It is debated if stimulant laxative abuse can cause loss of colon motility (“cathartic colon”); thus, osmotic laxatives are preferable for managing constipation.

Erosion of dental enamel through gastric acid is common in those who self‐induce vomiting on a regular basis. In these cases, NICE guidelines ⁵¹ encourage avoidance of brushing teeth immediately after vomiting, using non‐acid mouthwash after vomiting, and avoiding highly acidic foods and drinks. Regular dental reviews are also encouraged. The stomach acid‐induced erosion of oral mucosa and the pharynx can be associated with cough and dysphagia ¹⁵⁷ , while gastric acid in the oesophagus can induce reflux disease and rare mucosal laceration with episodes of hematemesis ¹⁵⁸ . Delaying, reducing or stopping self‐induced vomiting, together with medications to suppress acid production, are indicated in the presence of these symptoms.

Further complications associated with violent retching are subconjunctival haemorrhages and episodes of epistaxis. Indeed, recurrent epistaxis in young women without other medical causes may be a sign of covert BN, as well as hypokalaemia without ascertained causes ¹⁵⁹ . Rarely, vomiting may increase intrathoracic pressure causing pneumomediastinum, or may promote aspiration pneumonia. Of note, both vomiting and cessation of vomiting may cause parotid gland hypertrophy: sialagogues and anti‐inflammatory drugs are indicated in this case.

Persistent binge eating and BED are commonly associated with or lead to higher body weight or obesity. Individuals with BED in the general population report a variety of gastrointestinal symptoms, including dysphagia, acid reflux, bloating, abdominal pain, diarrhoea and constipation. In addition, respiratory (30%) and musculoskeletal (21%) problems are significantly increased in people with BED.

Individuals with BED – particularly due to obesity and increased risk of type 2 diabetes mellitus – have multiple risk factors for cancers. Other health concerns in these people include urinary incontinence and polycystic ovary syndrome. The latter is associated with insulin resistance and increased risk of infertility. Up to 23% of patients with this syndrome meet BED criteria ⁵ .

In a nationally representative study of US adults, the mean BMI of people with BED was 33.9 kg/m². Thus, health conditions commonly associated with BED include hypertension (31%), various heart conditions (17%), arthritis (24%), elevated cholesterol (27%) and triglycerides (15%), diabetes mellitus (14%), sleep problems (29%), general poor health, and metabolic syndrome ¹⁶⁰ .

All these findings highlight the need for a close monitoring of physical health in patients with EDs, who should be provided with full information about the consequences of starvation and purging and the ways to minimize risks.

ANTECEDENT AND CONCOMITANT PSYCHIATRIC CONDITIONS

Antecedent and concomitant mental disorders are exceedingly common in FEDs, but vary within and across clinical subtypes. This psychopathological complexity is influenced by biological and environmental factors ⁵ , ¹⁶¹ , ¹⁶² . For example, genome‐wide association studies (GWAS) in AN have found positive genetic correlations with OCD, major depression and anxiety disorders ¹⁶² , ¹⁶³ . These comorbid psychiatric conditions may influence treatment response of the ED. Meta‐analyses show poorer treatment outcomes, and higher dropout rates, for ED patients with greater comorbid psychopathology ¹¹⁷ .

Premorbidly, having a greater number of internalizing and externalizing behaviors in childhood has been linked to greater risk of developing an ED ¹⁶⁴ . Common premorbid psychiatric conditions are anxiety disorders, OCD, major depression, impulse control disorders, and obsessive‐compulsive personality traits ⁵ , ¹⁶⁵ , ¹⁶⁶ , ¹⁶⁷ , with the three years following onset of the first psychiatric disorder suggested as a key risk window for subsequent development of an ED ¹⁶⁶ , ¹⁶⁷ .

According to several systematic and meta‐analytic reviews and specific comorbidity studies ¹⁶⁶ , ¹⁶⁷ , ¹⁶⁸ , ¹⁶⁹ , the most common current and lifetime psychiatric comorbidities in AN are anxiety disorders (55‐59%), major depression (65‐81%), OCD and personality disorders (namely obsessive‐compulsive and borderline personality disorder, the latter often associated with the presence of self‐harm) ¹⁷⁰ , ¹⁷¹ . There is also notable co‐occurrence of AN and ARFID with autism ¹⁷² , with approximately 1 in 5 people with AN showing high autistic traits ¹⁷³ , and high food selectivity being very common in autistic children and adults ¹⁷⁴ .

Autistic patients with EDs have been found to have higher ED psychopathology, longer hospital stays, and increased depression and anxiety than non‐autistic ones ¹⁷⁵ , showing that this is a vulnerable population in need of appropriately tailored support. Assessment tools for autism are known to have poor sensitivity in girls and women, meaning that rates of autism in FEDs are likely to be underestimated.

The few studies reporting on comorbidities in men with AN have described similar findings as in females. However, in males there appears to be a higher prevalence of neurodevelopmental disorders – e.g., attention deficit hyperactivity disorder (ADHD) and autism – and substance use disorders compared to females ¹⁷⁶ .

Mood, anxiety and OCD symptoms may persist after recovery from an ED. For example, one study found that a quarter of patients who had recovered from AN met criteria for an anxiety or depressive disorder ¹⁷⁷ .

In BN, major depression is the most frequent comorbid psychiatric condition (72‐84%) ¹⁶⁶ , followed by anxiety disorders (56%), post‐traumatic stress disorder (PTSD), OCD, substance use disorders ¹⁷⁸ , and personality disorders (particularly borderline personality disorder) ¹⁷⁰ .

BED has a very similar profile of psychiatric comorbidity to BN, often being associated with lifetime comorbid mood disorders (70%), anxiety disorders, PTSD, substance abuse, personality disorders (mainly borderline personality disorder) and ADHD ⁵ , ¹⁷⁹ .

The concomitance of behavioral addictions is higher in BN and BED compared with restrictive‐type EDs, with compulsive buying (19%), kleptomania (18%) and pathological Internet use (12%) as the most frequently observed ¹⁷⁰ . In terms of gender, gambling is the most common behavioral addiction among men (16% of cases), while compulsive buying is the most common among women (17% of cases) ¹⁶⁷ .

Although there is limited evidence, “other” and subthreshold EDs (e.g., OSFED) have shown high levels of general psychopathology, and the most commonly described comorbidities are mood/anxiety disorders and substance misuse ¹⁶⁶ , ¹⁷⁰ .

Rates of lifetime non‐suicidal self‐injurious behavior and suicide attempts are higher in those with EDs than in other psychiatric disorders and in healthy controls ¹⁸⁰ . The former has been described in 27% of ED cases, and is more common in BN than in AN (33% and 22%, respectively) ¹⁸¹ . Suicide attempts appear equally present (about 22%) across ED subtypes ¹⁸² .

The presence of psychiatric comorbidity in EDs is often associated with greater severity of ED symptoms, more general psychopathology, maladaptive personality traits, greater cognitive impairment, longer duration of ED, and poorer prognosis ¹⁸ , ¹⁸³ , ¹⁸⁴ , ¹⁸⁵ , ¹⁸⁶ . Systematic reviews and meta‐analyses have found that psychiatric comorbidity is a significant predictor of relapse or treatment dropout ¹⁸ , ¹⁸⁵ , ¹⁸⁷ .

Screening for common psychiatric comorbidities can be done by the Psychiatric Diagnostic Screening Questionnaire (PDSQ) ¹⁸⁸ . For autism, the Autism Spectrum Quotient (AQ‐10) can be used as a screening tool ¹⁸⁹ . A careful psychiatric history will help establish the time course and potential interdependence of any comorbid disorders.

A protocol for how to best address co‐occurring mental health conditions in the treatment of EDs has been developed ¹⁹⁰ . This proposes that, where the comorbid condition appears to be a consequence of the ED, treatment can exclusively focus on the ED. For example, it has been clearly documented in starvation studies that low weight per se leads to low mood, anxiety and increased obsessionality ¹⁹¹ . Thus, in people with AN, weight restoration is likely to improve many of these symptoms. Specific intervention is advisable if a comorbid disorder is likely to impede engagement with ED therapy.

The most challenging cases are those in which the comorbid condition interacts with the ED and this impedes progress. In these cases, it may be possible to employ either concurrent or integrated interventions (e.g., modular approaches, or those that target transdiagnostic processes).

The Pathway for Eating disorders and Autism developed from Clinical Experience (PEACE) ¹⁹² is an example of how ED care may be tailored to take into account co‐occurring presenting features, in this case autism, with good results.

SOCIAL FUNCTIONING AND QUALITY OF LIFE

People with EDs often show impairment in interpersonal relationships, family function, work, finances, and social and private leisure activities ¹⁹³ . Early clinical and community‐based cross‐sectional studies reported poor social adjustment in people with EDs ¹⁹⁴ , and associations between lower adaptive function and greater illness severity ¹⁹⁵ . In the clinical context, assessment and monitoring of FED patients’ overall social functioning and quality of life, and the specific life domains affected by the disorder, may help them reflect on their values, improve motivation to change, and help refine treatment goals. It may also be a useful way of monitoring broad‐based progress over time.

Subjective generic measures, such as the WHO Brief Quality of Life Assessment Scale (WHOQOL‐BREF) ¹⁹⁶ , assessing quality of life as it relates to social interaction and perceptions of well‐being, are being increasingly applied to the assessment of EDs, particularly in recovery definitions ¹⁹⁷ , ¹⁹⁸ .

Health‐related quality of life measures assess people's appraisal of the impact of disease and treatment on their physical, psychological, social and somatic functioning and well‐being, and the relationships between severity of illness, functional status and disability ¹⁹⁹ , ²⁰⁰ . They have been argued to be the “primary endpoint” in clinical settings ²⁰¹ . Generic instruments – such as the 12‐Item Short Form Survey (SF‐12) ²⁰² and the EQ‐5D‐5L ²⁰³ – can be utilized across diagnostic groups, but several ED specific instruments are also now widely used – e.g., the Clinical Impairment Assessment scale (CIA) ²⁰⁴ and the Eating Disorder Quality of Life Scale ²⁰⁵ , ²⁰⁶ . These specific tools may be more sensitive at identifying differences between people with different levels of ED severity and/or different forms of an ED ²⁰⁷ , and have greater convergence with ED symptoms ²⁰⁸ .

An emerging field of assessment is the use of measures that allow individuals to report their person‐specific outcomes and unique concerns, such as the Psychological Outcome Profiles (PSYCHLOPS) ²⁰⁹ . Such instruments resonate with the broader conceptualization of recovery to encompass a personal perspective beyond symptom severity ¹⁹⁸ .

It has been observed that quality of life findings for people with AN may appear inconsistent, with a discrepancy between assessor (clinician) ratings of function and individual's symptom‐related quality of life ²¹⁰ . This has been ascribed to the ego‐syntonic nature of symptoms, such that an individual's sense of well‐being may be improved with the heightened sense of control and validation of successful weight loss in a society with high endorsement of the thin ideal. Nevertheless, using either generic or specific measures, a significant impairment of health‐related quality of life has been documented for all main EDs ²¹¹ . Further, poor quality of life has been reported in representative population studies of people with sub‐categories of OSFED, such as night eating syndrome ²¹² , ²¹³ , and in meta‐analyses of people with atypical AN ²¹⁴ and with ARFID ²¹⁵ , ²¹⁶ .

Empirical measurement is consistent with the lived experience of EDs as captured in qualitative research. In a qualitative meta‐synthesis of “severe and enduring” AN, functional impairment was reported as a “global impoverishment of self” in interpersonal relationships, physical health, mental health and socio‐economic functioning ²¹⁷ . This involved a series of losses, including intimacy and work/life functionality, which result in individuals living very lonely and isolated lives. Such experiences are not confined to AN, but occur to varying degrees across all EDs ²¹⁸ .

Research has also shown a wider impact of FEDs beyond the individual. Family and carer mental health impact and stress are amongst the highest for any psychiatric disorder ²¹⁹ . Three measures assessing caregiver burden, which can be easily applied in the clinical context, are the Experience of Caregiving Inventory ²²⁰ , the Eating Disorders Symptom Impact Scale (EDSIS) ²²¹ , and the Accommodation and Enabling Scale for Eating Disorders (AESED) ²²² . The EDSIS assesses carer social isolation, guilt, ability to manage nutrition in the family and cope with dysregulated behavior in the patient. The AESED evaluates how carers adapt to the illness.

One of the best‐known efforts to develop carer interventions to address these issues are the Maudsley Eating Disorders Collaborative Care Skills Workshops ²²³ , which use key elements of motivational interviewing and CBT to reduce caregiver distress and improve their skills in supporting their loved one. This model was developed originally for carers of adults with EDs and was found to be efficacious in reducing carer burden and expressed emotion (frequency of critical comments), and in improving self‐efficacy, skills and knowledge ²²⁴ . However, the approach also works well in families of adolescents with AN ²²⁵ , ²²⁶ .

NEUROCOGNITION

Several cognitive processes have been shown to be altered in people with EDs and are relevant to specific psychological and behavioral disturbances: reward processing, inhibitory control and decision‐making. Some processes (attention, working memory) have been found to be affected in ways that are largely attributable to starvation ²²⁷ . Several studies have found that cognitive impairment is a predictor of poor therapy response ²²⁸ , and might be reversible after ED recovery ²²⁹ .

Reward processes include hedonic value, motivational salience, and a set of learning processes in which the receipt of reward shapes behavior (reinforcement learning). The maladaptive behaviors that characterize EDs have been related to each of the above processes. People with AN, for example, have been shown to assign lower value to food, in general, compared with healthy peers – though whether this is a cause or a consequence of illness is unclear. People with BN or BED have been shown to have higher expectation of reward from food, but then to report decreased subjective experience of reward with receipt of food ²³⁰ .

Individuals with AN have been shown to be slower in learning from reward outcomes in laboratory tasks (reinforcement learning deficits) and to show abnormal brain responses to expected and unexpected reward outcomes ²³¹ . Some studies have found that reward learning differences between patients with AN and healthy peers have significance in clinical course ²²⁸ , ²²⁹ . Both behavior and brain research in AN suggest decreased reward properties of food and abnormalities in the processing of food value (which contributes to decision‐making) and abnormalities in dopamine functioning (dopamine is central to reward learning). The literature among patients with BN and BED is more limited, yet it does point to some differences in reward value in anticipation and receipt of food that may contribute to binge eating phenomena ⁵ , ²³² , ²³³ , ²³⁴ . Interestingly, patients with BED also show deficits in reinforcement learning (specifically, limited use of goal directed learning).

Cognitive control is a higher‐order executive function comprised of numerous cognitive functions. Components of cognitive control include inhibitory control, cognitive flexibility, and attentional control. This broad area has drawn a lot of attention in EDs because of the plausibility of a connection between cognitive control and eating behavior. Restrictive eating is likely reflective of an extreme inhibitory control. Restrictive eating also occurs in non‐binge meals for individuals with BN. At the same time, BN has also been associated with higher levels of impulsivity, which can be approached as a distinct neurocognitive process and is related to deficiencies in cognitive control.

Classic findings among people with AN include increased cognitive control during neuropsychological tasks, such as difficulty with set shifting (that is, changing response patterns when environmental contingencies change). Findings among patients with BN are more commonly suggestive of impaired inhibitory control – the inability to prevent a particular response ²³⁰ .

Decision‐making has been studied using monetary or food outcomes. The most commonly employed monetary tasks probe delay discounting, where participants choose between an amount of money available sooner or a larger amount available later. This complex task includes several sub‐components, and tends to identify that people with AN favour the larger‐later amount and patients with binge eating disorders tend to favour the smaller‐sooner ²³⁵ . In one set of food choice tasks, participants provide subjective ratings (healthiness and tastiness) as well as choice. This paradigm has identified differences in neural mechanisms of food choice between patients and controls, whereby people with AN show greater choice‐related activation in the anterior caudate and dorsal frontostriatal systems. While caution is warranted in making inferences about behavior based on brain activation patterns, these data are broadly consistent with habit‐centred models of AN ²³⁶ .

Assessment of neurocognitive impairments using task‐based measures is not typically feasible in clinical practice, due to their time‐consuming nature. Self‐report questionnaire measures exist for some of these constructs, e.g. cognitive flexibility, but with little or no correlation between questionnaire and task‐based measures ²³⁷ . Nonetheless, each of the neurocognitive domains reviewed has clinical implications in terms of identifying potentially new treatment targets. The reward deficits among patients with AN may make it more challenging to use learning in psychotherapy to change behavior in the service of health ²³⁸ . Novel psychological treatment approaches, such as positive affect treatment, aim to enhance attention to and appreciation of experience of reward to change behavior and are showing clinical promise in AN ²³⁹ . Further clarifying the ways in which people with AN have challenges in the reward and reward learning domains will help the development of these treatments.

Habit strength of illness‐related routines, e.g. in relation to eating, in AN is related to greater illness duration and severity ²⁴⁰ , and multimodal interventions targeting habits have shown promise and are the subject of further investigation ²⁴¹ , ²⁴² . Treatment designed to directly target neurocognition (e.g., cognitive set‐shifting weakness and overly detail‐focused thinking), namely cognitive remediation therapy for AN, has not as yet shown any clear advantages over different control treatments in improving neurocognition or other outcomes. However, it may reduce dropout ²⁴³ , ²⁴⁴ .

Among individuals with BN, psychotherapy may need to specifically focus on enhancing inhibitory control. Neurocognitive trainings with this target have indeed shown promise ²⁴⁵ , ²⁴⁶ , ²⁴⁷ , and are being explored with some success also in BED ²⁴⁸ , ²⁴⁹ . Computational psychiatry has potential to provide a more granular evaluation of these neurocognitive processes, and to identify latent variables (i.e., aspects of decision‐making that are not directly observable) that will be useful for personalization of treatment ²⁵⁰ .

SOCIAL COGNITION AND EMOTION

Most research on social cognition has been conducted in people with AN ²⁵¹ , ²⁵² . The findings indicate impairments across a range of domains ²⁵¹ , ²⁵² , ²⁵³ , ²⁵⁴ , ²⁵⁵ , including communication; affiliation‐related outcomes, as reflected by self‐reported insecure attachment ²⁵¹ ; and a tendency towards negative interpretation of social scenes ²⁵² . People with AN have also shown deficits in emotion processing, including alterations in retrieval of emotions, startle response, pleasure ratings to affective touch, and emotional expression ²⁵¹ , ²⁵² , ²⁵⁵ .

Impairments in domains related to the evaluation of (emotional/cognitive) states of others have been expressed in (facial) emotion recognition difficulties ²⁵¹ , ²⁵² , ²⁵³ , ²⁵⁴ . Theory of Mind (ToM) impairments in people with AN were reported in earlier meta‐analyses ²⁵¹ , ²⁵³ , but more recent evidence is inconsistent ²⁵⁵ . ToM outcomes in AN seem to be influenced by the task used, and previous studies mainly relied on a single task which might assess emotion recognition rather than ToM ²⁵⁵ .

There has been less research on social cognition and emotion in BN and BED, rendering it difficult to draw conclusions about these disorders ²⁵¹ , ²⁵² , ²⁵⁶ . Studies have centred on emotion‐related domains, resulting in some evidence for difficulties in emotion processing among patients with BN, and self‐reported emotion regulation difficulties among patients with BN or BED ²⁵² , ²⁵³ . Recent meta‐analytic data outline the transdiagnostic character of emotion regulation difficulties, indicating strong relationships between maladaptive emotion regulation strategies such as rumination, avoidance or suppression of emotions and ED symptom severity ²⁵⁷ .

Whilst task‐based measures of social cognition are too cumbersome for use in clinical practice, brief questionnaire‐based measures exist that tap into emotion (dys)regulation. Two widely used tools are the Emotion Regulation Questionnaire (ERQ) ²⁵⁸ , which assesses cognitive reappraisal and expressive suppression, and the Difficulties in Emotion Regulation Scale (DERS) ²⁵⁹ , which assesses a broader range of facets of emotion (dys)regulation.

In terms of clinical implications, the cognitive‐interpersonal maintenance model of AN posits that deficits in socio‐emotional functioning contribute to development and persistence of illness, and existing data from patients are largely consistent with this model ³⁵ , ²⁶⁰ . The MANTRA approach is derived from this maintenance model ³⁵ , ²⁶⁰ and includes interventions to address the emotional and social life of patients and work with close others to strengthen interpersonal functioning.

For binge‐like EDs, despite the sparsity of mechanism research, disease models and treatment approaches are increasing. Current efficacy data for BED treatment match very well with the evidence on socio‐emotional deficits, as interpersonal psychotherapy and dialectical behavior therapy have shown efficacy in randomized controlled trials alongside CBT‐ED ⁵ . Interpersonal psychotherapy specifically focuses on interpersonal functioning, communication and relationships, while dialectical behavior therapy predominantly addresses emotion regulation skills, and has a focus on interpersonal issues.

Finally, the reported abnormalities in the social brain networks in people with EDs ²⁵⁶ suggest that neuromodulatory approaches may have potential, with a putative treatment focus on improving social functioning and emotion regulation by targeting associated brain areas by neurofeedback approaches or non‐invasive brain stimulation ²⁶¹ .

DYSFUNCTIONAL COGNITIVE SCHEMATA

Cognitive theories propose that the development and maintenance of psychopathology can be partially attributed to processing disorder‐salient stimuli preferentially above other information types ²⁶² . Attention bias refers to the preferential processing of salient stimuli such that the focus of attention influences responses ²⁶³ . Techniques targeting attentional bias (e.g., dot‐probe, Stroop, free recall, eye‐tracking) aim to manipulate selective attention for disorder‐salient information. Various cognitive models share the premise that biased patterns of basic information processing, operating early within the cognitive system and at a low level, play a central causal role in vulnerability to experience intense emotional symptoms ²⁶⁴ .

Evidence suggests that people with EDs experience greater bias away from food stimuli, and bias towards body‐related stimuli, compared to healthy controls ²⁶⁴ . The Stroop task is the most widely used measure of attentional bias in EDs, with results suggesting that attention bias to food stimuli is comparable in people with AN and restrained eaters, but greatest in those with BN ²⁶⁵ and BED ²⁶⁶ .

Patients with EDs are more likely to attribute negative body interpretations to ambiguous sentences and scenarios compared to healthy individuals ²⁶⁴ . Attentional, interpretation and memory biases for stimuli pertaining to negative self‐worth have been implicated in the development and maintenance of EDs. People with EDs experience elevated sensitivity hypothesized to be triggered by a negative interpretation bias, the tendency to interpret ambiguous social situations negatively and to anticipate negative endings ²⁶⁷ .

Cognitive bias modification (CBM) refers to a class of interventions targeting cognitive processes considered key in the etiology and maintenance of different psychopathologies. Research has focused primarily on two types of CBM: attention bias modification (ABM) and cognitive bias modification for interpretation (CBM‐I). Bias modification trains participants to attend to neutral or positive stimuli in preference to negative, threatening stimuli. There is weak evidence of the efficacy of ABM for appetitive behaviors ²⁶⁸ , with reduced attentional avoidance of food stimuli and a reduction in ED symptoms ²⁴⁵ .

CBM strategies have targeted many aspects of ED psychopathology, including concerns about appearance and self‐worth, with moderate‐to‐large reductions in bias, though with smaller and less consistent effects on symptomatology ²⁶² . CBM has also been shown to decrease negative interpretation bias towards ambiguous social situations, with strong effects in women with AN and BN ²⁶⁹ , ²⁷⁰ . CBM could provide a useful treatment enhancement by increasing sensitivity to positive social feedback and reducing sensitivity to social criticism from family and peers. The online nature of the training may appeal to younger populations. It is possible that in future these strategies offer the potential to augment treatment‐as‐usual strategies.

Early maladaptive schemata are broad, pervasive and dysfunctional belief systems regarding oneself, others and the world, which develop during childhood and impact functioning. They can be assessed by the Young Schema Questionnaire ²⁷¹ , which includes eighteen such schemata across five domains: disconnection/rejection, impaired autonomy/performance, impaired limits, other directedness, and over‐vigilance/inhibition. Compared to healthy controls and other clinical populations, individuals with EDs score higher on most schemata ²⁷² .

Schema therapy was developed to address the causes of early maladaptive schemata and the impact of these on present‐day functioning. A recent systematic review of schema therapy in EDs ²⁷³ found four articles (including one RCT ²⁷⁴ ) with 151 participants that met inclusion criteria. In the RCT, schema therapy performed comparably to CBT and appetite‐focused CBT for people with binge‐eating ²⁷⁴ . Given the small number of studies, few firm conclusions can be drawn.

PERSONALITY TRAITS

Recent studies report similarities in personality traits across ED diagnoses. For example, a comprehensive review suggested that perfectionism, neuroticism and avoidance motivation are elevated, while extraversion and self‐directness are reduced, in AN, BN and BED ²⁷⁵ . Despite these similarities, some ED diagnostic differences did emerge across certain personality traits. For example, there was considerable evidence that impulsivity was higher in BN than in AN, but further analysis of AN subtypes revealed that individuals with AN who binge and purge showed levels of impulsivity that are roughly equal to BN, and notably higher than in the restricting AN subtype.

Cluster analytic studies and latent structure models have been used to try and identify personality‐based groups within ED diagnoses ²⁷⁶ , ²⁷⁷ , ²⁷⁸ , ²⁷⁹ . Using a variety of personality measures, and different statistical approaches, three personality‐based subtypes were identified across studies for AN and BN: under‐controlled, over‐controlled, and low psychopathology ²⁸⁰ . When these groups were compared, they showed significant and clinically relevant differences in patterns of treatment utilization and response, psychosocial functioning, and history of various etiological factors ²⁷⁸ , ²⁸¹ .

Another important area of study is the longitudinal predictive value of personality traits in the clinical course of an ED. Studies which have assessed personality functioning at baseline, and used it to predict disorder course and outcome, suggest that elevations in emotion dysregulation and impulsivity predict a negative ED outcome ²⁸² , ²⁸³ . However, there has also been evidence suggesting that ED symptoms are correlated with elevations in personality traits, and that personality trait elevations subside as ED symptoms improve ²⁸⁴ , ²⁸⁵ .

When personality traits and ED status are both assessed repeatedly in longitudinal studies, findings are mixed. One study found that personality functioning had no significant influence on ED outcomes over a five‐year time frame in individuals with BN ²⁸⁶ , while a 17‐year follow‐up of a transdiagnostic sample of ED individuals found that elevated borderline personality scores at baseline predicted a more negative ED course, but that changes in personality traits or ED symptoms did not significantly influence the other condition ²⁸⁷ . Thus, the longitudinal relationship between personality traits and ED symptoms remains relatively unclear, and more robust, prospective longitudinal studies are needed to test this relationship.

It is important to consider whether personality traits may moderate the effectiveness of ED treatments. Studies in this domain are relatively limited and have revealed a pattern of inconsistent findings, depending on the ED diagnosis in the study and the types of treatment implemented. A literature review which included seven RCTs and four treatment‐related naturalistic follow‐up studies concluded that personality traits typically had some type of impact on treatment outcomes, but this varied across studies ²⁸⁸ . For example, treatment outcomes for BN did not seem to be significantly impacted by the level of impulsive or emotionally dysregulated personality traits ²⁸⁹ , ²⁹⁰ , whereas heightened avoidance and inhibition personality traits displayed a negative impact on outcomes for BED treatment and also an increased likelihood of attrition in AN studies ¹⁸³ .

Integrative cognitive affective therapy (ICAT) has been evaluated as an emotionally‐focused ED treatment, and may be helpful for individuals with personality disorder features. In a comparison of ICAT and CBT‐ED for BN, the two treatments did not differ overall in their impact on bulimic symptoms, depression and anxiety ²⁹¹ . However, when specific personality traits were included in the statistical model to examine differential efficacy of each treatment for different personality types, there were differences. Specifically, individuals higher in stimulus seeking had greater reductions in bulimic behavior and ED psychopathology when receiving ICAT than when treated with CBT‐ED, whereas individuals lower in stimulus seeking had greater reductions in bulimic behavior with CBT‐ED than ICAT. Additionally, individuals with higher affect dysregulation had greater reductions in ED psychopathology in the ICAT than in the CBT‐ED condition ²⁹² .

For clinicians wishing to evaluate personality, brief assessment tools for ICD‐11 and DSM‐5 personality trait domains include the 17‐item Personality Assessment Questionnaire for ICD‐11 (PAQ‐11) ²⁹³ and the Personality Inventory for DSM‐5‐Brief Format (PID‐5‐BF) ²⁹⁴ , respectively.

FAMILY HISTORY

There is substantial evidence that EDs run in families. Ascertaining family history is therefore an important element of clinical assessment.

Family and twin studies show that heritability estimates are high across EDs, with variable estimates depending on the disorder ²⁹⁵ . Twin‐based heritability estimates are highest for AN (0.28‐0.74) ²⁹⁶ , ²⁹⁷ , ²⁹⁸ ; intermediate (0.55‐0.62) for BN ²⁹⁷ , ²⁹⁹ , ³⁰⁰ ; and lower for BED (0.39‐0.45) ³⁰¹ , ³⁰² , ³⁰³ . A high twin‐based heritability of a broad ARFID phenotype (0.79) has also been identified ³⁰⁴ .

The risk of EDs in first‐degree relatives of individuals with EDs is about 7‐10‐fold higher compared to the general population ³⁰⁵ . An investigation of whole population samples in Denmark and Sweden estimates heritability of diagnosed AN at 0.36 and BN at 0.39. This study also found that having a parent with AN is associated with a 3‐fold increased risk of being diagnosed with AN ³⁰⁶ . Family history of other psychiatric disorders, such as anxiety and depression, is also higher in individuals with EDs ³⁰⁷ .

Ascertaining parental ED and other psychiatric morbidity might be helpful in aiding treatment in children and adolescents (as that morbidity might impact on treatment), but also in identifying potential traits contributing to individual clinical presentations (e.g., in the case of autism spectrum disorder or anxiety) and formulation, as well as aiding individual understanding of the disorder itself.

The last ten years have seen an exponential growth in our understanding of genetic risk for EDs (particularly AN). Two GWAS have been carried out for AN ³⁰⁸ , ³⁰⁹ , providing insights into the genetic etiology of AN and its genetic overlap with other psychiatric disorders (e.g., OCD, schizophrenia, anxiety) and anthropometric/metabolic factors (e.g., BMI, insulin levels, diabetes mellitus).

Whilst quantifying genetic risk (in the form of polygenic risk scores) is not yet useful in clinical practice, naming the contribution of genetic factors to the development of EDs in a clinical setting may be helpful. Clarification of the role of genes as well as environment during a diagnostic assessment might allow not only moving away from the dichotomy of individual responsibility in illness development (which often manifests as guilt in adult individuals and in caregivers) vs. genetic determinism, but can also help defining treatment targets and adherence to treatment ³¹⁰ , ³¹¹ , ³¹² .

ENVIRONMENTAL EXPOSURES

Most EDs emerge before the age of 25 years. Of these, 40% emerge in adolescence and 49% in early adulthood ³¹³ . When considering environmental exposures, the notions of intersectionality and developmental sensitivity are critical. For example, genetic influences and non‐shared environmental influences impacting on the emergence of disordered eating increase significantly over puberty, while shared environmental influences decrease and become negligible ³¹⁴ . In other words, there is dynamic interplay between multiple genetic and non‐shared environmental risk factors over the critical developmental span where EDs emerge.

Therefore, none of the early or recent environmental exposures discussed in this section should be viewed as silos, but rather as part of a rich and complex tapestry of risk and maintaining factors. A comprehensive assessment of relevant environmental exposures in practice should consider these interactions and the unique constellations of factors for each patient. It is notable that the vast majority of studies on environmental exposures is focused on general ED pathology or BN, and to a lesser extent AN and BED. Relatively little is known about risk factors for other FEDs, including ARFID, pica and rumination disorder.

Early environmental exposures

A range of early environmental exposures have been investigated in relation to EDs, and have been found to contribute to their development, although much of the research relies on retrospective studies, potentially impacting the reliability of the evidence generated.

Available evidence from large studies on the role of pregnancy, obstetric and perinatal factors points to higher risk related to prematurity, lower birth‐weight, and small for gestational age for AN, higher birth weight/large for gestational age for BED, and pregnancy smoking and prematurity for BN ¹⁶³ , ³¹⁵ . Although few studies are available in ARFID, initial findings suggest higher prevalence of preterm birth, postnatal complications and invasive procedures (involving the gastrointestinal or respiratory tract) postnatally in children with this condition ³¹⁶ . This matches clinical observations and has relevance for guiding treatment in ARFID.

Enquiring about obstetric complications in the context of EDs may not impact on treatment planning or prognosis of the individual. However, where women with an ED are planning to get pregnant, it is important to provide them with information on reducing ED behaviors during pregnancy, as this may improve outcomes for their unborn child and decrease the risk of intergenerational perpetuation of the ED. In ARFID, obtaining a detailed history of perinatal complications is essential, as the sequelae of obstetric factors and/or early postnatal invasive procedures may have impacted on oral sensitivity and/or led to food aversion(s) and therefore may require a specific focus in treatment ³¹⁶ , ³¹⁷ .

The literature supports an association between childhood maltreatment and EDs. An umbrella review ³¹⁸ found associations between childhood sexual abuse and BN, and between appearance‐related teasing and any ED, based on meta‐analyses pooling longitudinal observational studies. These findings corroborate the role of early traumatic experiences as risk factors for EDs and add to previous evidence of higher prevalence of childhood maltreatment in EDs compared to both healthy controls and individuals with other psychiatric disorders ³¹⁹ .

Compared to patients without a history of maltreatment, those with such a history have a more severe clinical presentation, earlier onset, higher rates of comorbidity ³¹⁹ and poorer treatment response ³²⁰ , ³²¹ . Putative mediators of the association between childhood maltreatment and EDs have been investigated through different methodologies ³²² , ³²³ , but results are not conclusive ³²⁴ . Preliminary experimental evidence suggests that childhood maltreatment may lead to a heightened sensitivity to social stress in adulthood ³²⁵ .

Overall, early trauma may be considered an important diagnostic specifier that can alter treatment response, and the evaluation of childhood maltreatment history should be part of clinical routine assessment. The Childhood Trauma Questionnaire ³²⁶ is a self‐report tool commonly used to assess childhood maltreatment in terms of emotional abuse and neglect, physical abuse and neglect, and sexual abuse. However, it does not provide information about timing, severity or duration of maltreatment exposure and relies on the individual's recall. Bearing in mind these possible biases, this instrument can be considered for use in clinical practice.

There is preliminary evidence for the use of schema therapy ²⁷⁴ and cognitive analytical therapy (CAT) ³²⁷ in the treatment of EDs, and established evidence for dialectical behavior therapy in the treatment of BED ⁵ . These treatment models consider early trauma experiences and may provide alternative options for patients who have not been helped by first‐line ED treatments (e.g., CBT‐ED, MANTRA). However, first‐line evidence‐based treatments should always be tried first, and these can be tailored to ensure that care is trauma‐informed and considers comorbidities, including PTSD, if present ¹⁹⁰ .

Early adverse experiences also include insecure attachment bonds. Disrupted interactions with early caregivers promote the development of maladaptive schemata about the perception of self and others ³²⁸ . Insecure attachment is more common in individuals with EDs compared to community controls, and meta‐analytic evidence of this association points to medium‐to‐high effect size, despite some limitations ²⁵¹ , ³²⁹ . Maladaptive emotion regulation and depressive symptoms were the strongest mediators of this relationship ³³⁰ . However, given a lack of longitudinal and experimental studies, these findings are not exhaustive. Moreover, preliminary data suggest a possible reciprocal interaction between insecure attachment and childhood maltreatment ³³¹ , ³³² .

Of note, insecure attachment can affect treatment outcome via its effect on the development of the therapeutic alliance ³²⁹ . Several self‐report measures have been developed to assess attachment style/relationships, including the Experience in Close Relationship ³³³ and the Attachment Style Questionnaire (ASQ) ³³⁴ . Focal psychodynamic therapy ³³⁵ is an evidence‐based treatment option for AN which targets attachment‐related issues.

Recent environmental exposures

The role of stressful life events in precipitating ED onset is well established, with interpersonal and sexual‐type events being of particular significance ³³⁶ . There is also some evidence from qualitative and quantitative studies that negative life events are implicated in ED relapse and poorer treatment outcome, whereas positive life events may facilitate recovery ³³⁶ .

The wide range of interview and questionnaire measures to assess life events and difficulties have been previously reviewed in this journal ³⁰ , ³¹ , ³² , ³³ . A novel electronic, structured approach for obtaining and graphically representing data on life stresses and their impact on mental health is provided via the Tulsa Life Chart (TLC) ³³⁷ . This tool covers distinct life epochs (birth to elementary school, elementary school, middle school, high school, young adulthood, age 25 to 35, 35 to 45, etc.) and assesses negative and positive life events, and epoch‐ and event‐related mood ratings, in the broader context of factors such as school attendance, hobbies, jobs, social support, substance use, mental health treatment, and family structure changes. As yet, this is an interview‐based tool, but self‐report versions are planned.

The importance of stress in the onset and course of EDs has been highlighted by the significant increases in ED presentations at treatment facilities around the world since the advent of the COVID‐19 pandemic ³³⁸ . Young people seemed particularly affected. Multiple acute and chronic stressors were identified as contributing to worsen the mental health of young people during the pandemic, such as social isolation, excessive screen time and social media use, parental stress, poor parent‐child relationship, low socioeconomic status, pre‐existing mental health conditions and/or disabilities ³³⁹ , ³⁴⁰ . It can be helpful for clinicians to enquire about how FED symptoms were impacted by, or developed alongside, the COVID‐19 pandemic, in order to understand any precipitating or maintaining factors relevant to treatment.

Strong evidence exists to implicate appearance pressures in the development of EDs ³⁴¹ , ³⁴² , and preventive programmes targeting thin ideal internalization have been shown to reduce incidence of EDs ³⁴³ . A longitudinal study of risk factors for the emergence of disordered eating in adolescent females showed that weight‐related peer teasing significantly strengthened genetic risk for the development of disordered eating, a gene‐environment interaction ³⁴⁴ . A meta‐analysis of experimental evidence showed that exposure to social media appearance‐ideal images had a moderate negative effect on body image in the short term, although longer‐term impacts of naturalistic use of social media are less clear ³⁴⁵ , ³⁴⁶ .

Given the significant pressures around appearance on many patients with EDs, enquiring about appearance ideals as well as patient's social media use is an essential part of the clinical assessment. The Sociocultural Attitudes Towards Appearance Questionnaire‐4‐Revised (SATAQ‐4R) ³⁴⁷ is a widely used measure of appearance pressures, but is relatively long to use in routine clinical practice.

As mentioned earlier in this paper, EDs affect certain minoritized groups at elevated rates. These include people with a second‐generation migration background ³⁴⁸ ; LGBTQ+ individuals ³⁴⁹ , as well as people at higher body weight. There are consistent correlations between experiences of discrimination (racial, sexual, ethnic, weight‐based) and disordered eating symptoms, suggesting that minority‐stress factors may partially explain this higher prevalence ³⁵⁰ . Individuals with BN and BED are two to three times more likely to have been bullied or teased about their appearance compared to those without EDs ³⁵¹ .

A short measure of everyday discrimination which is able to accommodate different minoritized characteristics, asking people if they are being discriminated in their daily life (e.g., whether they are treated with less courtesy or respect or whether they are given a poorer service) and how often this happens, may offer a useful way of starting a conversation about the topic of minority stress ³⁵² .

Income level and employment status do not predict the development of ED ³⁴² . However, there is some evidence that socioeconomic status can be associated with the type of ED behaviors: low socioeconomic levels may predict binge‐purging behaviors, while higher education may be associated with food restriction ³⁵³ . More recent work has shown that intersectionality is important here, with some evidence of increased likelihood of a positive ED screen in groups with lower socioeconomic status, particularly those from Hispanic/Latinx and sexual minority backgrounds ⁹⁷ . Food insecurity may be one factor driving EDs in groups with lower socioeconomic status, potentially via a “feast and famine” cycle of food availability.

Nascent evidence shows that bulimic‐type difficulties (binge eating and compensatory behaviors) are more common in adults (but not necessarily adolescents) who are food insecure ³⁵⁴ , ³⁵⁵ . Moreover, experience of childhood food insecurity is associated with binge eating behaviors ³⁵⁵ . Importantly, low socioeconomic status is a barrier to early access to specialist treatment for EDs ¹⁶¹ .

The US Department of Agriculture offers a suite of measures to assess food security in adolescents, adults and households, with a brief, 6‐item screening tool typically being most suitable for use in clinical practice ³⁵⁶ . It is important that, where food insecurity is identified, the individual's lack of financial resources is taken into account in meal planning and other aspects of treatment (e.g., travel).

STIGMA

Stigma is consistently found to have a major and adverse impact on people's well‐being, and their ability to access and engage in appropriate treatment for their ED. For people with an ED, stigma occurs across many levels and is multifaceted. It is experienced regarding the presence of a mental disorder, the presence of an ED, and the presence of high or low weight. It may be expressed in cognitive (e.g., stereotypic beliefs that people with AN should “just eat”), emotional (e.g., disgust towards people who vomit after eating), and behavioral (e.g., not providing appropriate care for someone who is of “adequate” weight) terms ³⁵⁷ . Stigma is also typically suffused with moral connotations, i.e. labelling ED symptoms and behaviors as “sinful or selfish” and arising from “vanity” or “greed”, or being “wasteful of food”.

The experience of stigma may subsequently be internalized. A negative association was found between self‐stigmatization and effective help‐seeking ³⁵⁸ , and it was observed that this may be a stronger phenomenon for men than women with EDs ³⁵⁹ . However, a meta‐synthesis of 29 papers ³⁵⁸ also highlighted people's capacity to utilize “resistive” strategies to challenge stigma, e.g. seeking validation through peers, and reported that such strategies are associated with better outcomes.

A large body of literature has affirmed the association between stigma and inadequate treatment, although this mostly lies in the qualitative research space ³⁶⁰ . Many investigations have described a fear of exposure and an anticipated or actual negative response from health professionals when seeking treatment, albeit in one meta‐analysis there was a suggestion that the role of stigma may have been overstated as a barrier to care ³⁶¹ .

Stigma and beliefs that EDs are not serious or impactful may have contributed to their under‐recognition and neglect from policy makers and funders of mental health training, research and treatment ³⁶² , ³⁶³ .

The modified Weight Bias Internalization scale (WBI‐M) ³⁶⁴ can be used by patients across different body sizes to assess their weight‐related self‐stigmatization. The patient‐rated Scale for Treatment‐Based Experiences of Weight Stigma (STEW) ³⁶⁵ may help assess whether individual therapists, treatment teams, or peers engage in stigmatizing beliefs or actions in relation to people with EDs at higher body weight.

At a population level, improved health literacy and understanding should reduce stigma. However, this is difficult to achieve. Training of health practitioners to reduce bias and combat/disrupt the impact of stigma and shame around help‐seeking is also needed. The importance of moving research forward has been emphasized ³⁵⁷ , not least as it is the vulnerable (young and socio‐economically impoverished people) who experience most of the negative impacts of stigma.

PROTECTIVE FACTORS

Potential protective factors – which may reduce ED risk, mitigate the noxious influences of recent and early environmental exposures, and support recovery – include some ED‐specific factors (such as learning to appreciate or accept one's body, challenging unrealistic appearance ideals, minimizing exposure to appearance‐focused material in social media, learning to enjoy a healthy diet) and some broader factors (such as enhancing self‐compassion to offset the self‐criticism that can be a consequence of unfavorable comparisons, and strengthening social connection).

Body appreciation is defined as accepting, holding positive attitudes toward, and valuing the body. A meta‐analysis of 240 studies found that body appreciation was inversely associated with several indices of eating and body image disturbances, as well as general psychopathology (depression, anxiety), and positively associated with several well‐being constructs (such as self‐esteem and self‐compassion) ³⁶⁶ . A longitudinal study of 3,039 women found that body appreciation was associated with lower levels of eating pathology eight months later ³⁶⁷ . The Body Appreciation Scale‐2 (BAS‐2) ³⁶⁸ is a widely used 10‐item measure to capture body appreciation in adults, and a recent 2‐item version offers a brief measure that may be particularly practical for clinical use ³⁶⁹ .

There is current debate in the literature about positive body image (an individual's ability to conceptualize her/his body with love, respect and appreciation) ³⁷⁰ versus body neutrality (a neutral and mindful attitude toward the body, with self‐worth being broadly defined and not dominated by appearance) ³⁷¹ . Future research is required that examines meaningful differences and impacts of the two concepts. Currently, body‐image focused treatment components or adjuncts to treatment for FEDs, such as mirror or virtual reality‐based exposure treatments ³⁷² , tend to focus more on reducing body dissatisfaction and achieving a neutral or accepting stance to one's body rather than increasing appreciation.

Approaches that enhance individual critical appraisal of appearance ideals are effective at protecting against or reducing body image concerns and, to some extent, in helping FED treatment. These include enhancing media literacy (the ability to critically engage with and evaluate media content), as well as directly challenging thin ideal internalization using cognitive dissonance‐based exercises. These approaches support individuals in being able to critically consider the credibility of media sources, the veracity of images, and the values and intentions that underpin media content creation, thus minimizing internalization of appearance ideals and engagement in social comparisons ³⁷³ .

Public health approaches to tackle unhelpful media content are an important element. However, attempts made to minimize the negative impact of media content, via strategies such as disclaimers on images (e.g., highlighting when digital manipulation has been applied), have consistently been shown to be ineffective ³⁷⁴ .

Self‐compassion is increasingly recognized as a factor aiding ED recovery. It involves treating oneself with kindness and encouragement rather than self‐criticism when facing challenges. It has been hypothesized to be protective by providing an adaptive process to manage negative emotions, self‐criticism and self‐critical perfectionism which might otherwise trigger/exacerbate ED behaviors ³⁷⁵ .

A meta‐analysis found that self‐compassion is associated with lower eating pathology, and that self‐compassion interventions are effective at reducing ED symptoms ³⁷⁶ . In the clinical context, the Self‐Compassion Scale ³⁷⁷ is useful as a brief assessment tool. Most empirically‐supported ED treatments, with the exception of MANTRA ³⁷⁸ , do not directly target improving self‐compassion. Compassion‐focused therapy may be particularly useful for patients with a trauma history ³⁷⁹ .

High levels of loneliness and disordered eating have been shown to be mutually reinforcing ³⁸⁰ , suggesting that building positive social connection may be a key protective approach. Analysis of the National Longitudinal Study of Adolescent to Adult Health found that both mother‐ and father‐connectedness is associated with lower odds of onset of a range of ED symptoms six years later in girls (but not in boys) ³⁸¹ . Parents can also build a protective home environment by supporting regular family meals and engaging in conversations around food that do not focus on dieting or weight ³⁴² .

Relatively little is known about the protective role of positive social connection to peers, but there is some evidence that a strong sense of social support from friends may buffer against the development of body image concerns and disordered eating ³⁸² . Importantly, positive social support is also the most prominent facilitator to engaging in help‐seeking for FEDs ³⁶⁰ .

Loneliness can be measured directly (i.e., by asking “how often do you feel lonely?”) or indirectly (i.e., by asking about emotions associated with loneliness). The 3‐item UCLA Loneliness Scale is the most widely used measure of loneliness, assessing relational connectedness, social connectedness, and self‐perceived isolation ³⁸³ . The 10‐item Significant Others Scale (SOS) ³⁸⁴ allows differentiation between actually received and ideal level of emotional and practical social support from others. The 19‐item Medical Outcomes Study Social Support Survey ³⁸⁵ distinguishes between emotional, tangible and affectionate support and positive social interaction with others.

Given the evidence that dieting is a risk factor for EDs ³⁴¹ , it seems likely that good nutrition may protect against the development of or relapse from an ED. Whilst most Western countries have published official guides for healthy eating, only the Australian REAL Food Guide is tailored to the specific needs and beliefs of people with FEDs, and includes a user‐friendly food pyramid ³⁸⁶ .

A thorough clinical assessment should include evaluation of ED‐specific and broader protective factors that may support improvements or recovery from an ED. In addition to covering the areas described above, it will be helpful to ask patients to complete the VIA Assessment Suite for Adults ³⁸⁷ , to help them attend to those strengths that may have become neglected or side‐lined during their illness, and those that they might not recognize as having the potential for supporting their recovery.

DISCUSSION

This paper has attempted to systematically describe some of the key clinical domains that should be considered when trying to personalize the management of FEDs, along with relevant measurement tools. For most of the domains presented here, it is currently not known how to prioritize, combine or sequence these considerations in order to improve patient outcomes, except for the need to prioritize high medical risk when determining treatment setting and intensity.

At the heart of FED management is psychological treatment. A generic three‐dimensional conceptual framework of personalization of psychological therapies has been proposed ³⁸⁸ . This framework includes the timing at which personalization decisions are made in a patient's treatment pathway – e.g., at initial assessment, throughout the course of treatment, or at the end of it (such as to determine need for and type of relapse prevention). Secondly, it considers the level, type and intensity of treatment, choice, combination and sequencing of techniques, or style of delivery. Lastly, it deals with what the authors call “structure”, i.e., the method of personalization that is used. The latter is on a continuum from informal idiosyncratic personalization, based on clinician “intuition” and/or patient choice, through to theory‐ and data‐informed integrative or matching approaches and other adaptations of standard treatments for particular populations, up to data‐driven actuarial approaches to enhance treatment decisions. Evidence‐based clinical guidelines lie somewhere in the middle of the “structure” spectrum. Whilst this framework was designed specifically with psychological therapies in mind, it can readily be applied more broadly to other treatment and clinical management decisions.

Across the EDs, there are very few replicable baseline predictors of outcome ¹¹⁷ . A recent meta‐analysis ¹⁸ found that children and adolescents (i.e., those with typically shorter illness duration) had more favorable outcomes than adults, and those with self‐harm had poorer outcomes than those without. These data support the principles of illness staging and emphasize the importance of early detection and easily accessible early intervention with developmentally and illness stage appropriate treatments. The data also highlight the need to take comorbidities into account when personalizing treatment for FEDs.

A key finding is that, across different evidence‐based treatments and all EDs, early treatment response (i.e., reliable symptom improvement during the first four sessions) is a solid predictor of recovery ⁴² . In AN, trajectory studies based on BMI during early treatment sessions found that there were broadly three groups in terms of outcomes: “hares”, “tortoises” and non‐responders ³⁸⁹ , ³⁹⁰ . These studies confirm that those with the most rapid early change do best in the longer term (“hares”), but also identified a class of patients that after a slower start did well in the longer term (“tortoises”), together with another group that deteriorated or was unchanged.

Thinking about types and levels of interventions in different populations, for those at the milder end of illness severity or with fewer comorbidities and complexities, there is growing interest in programme‐led, focused interventions ³⁹¹ and single‐session interventions ³⁹² , which may increase access to treatment, reduce waiting times, and allow treatment to be provided in more flexible and personalized ways (including online and digital solutions). Conversely, there is a need for further research on intensive treatment programmes for EDs and how to optimally use these for patients who require higher levels of support and do not benefit from standard outpatient care ³⁹³ .

Personalization through adaptation of Western evidence‐based treatments to different cultures – e.g., East Asia ³⁹⁴ – or to minoritized ethnic/racial groups ³⁹⁵ is also an area of growing interest, but with limited research. It is also important to hold in mind that, in the case of these groups, minority status is often intersectional with other indicators of disadvantage which may need to be considered. It has been posited that current measures may not adequately capture eating pathology in marginalized groups, given the different drivers of pathology in such groups (e.g., dietary restraint due to food insecurity rather than to weight/shape concerns). Biased measurement may contribute to sub‐optimal diagnostic screening, prevention and outcome measurement ³⁹⁶ .

Personalization based on age fits with the traditional divide into child/adolescent and adult services, and some staging approaches. However, emerging evidence suggests that family‐focused and individual therapies, previously recommended for young people aged below 18 years or adults respectively, can be effectively developmentally adapted to fit older or younger age groups ¹⁷ , ³⁹⁷ , by including more or less intensive family involvement and support from others (partners, friends), whilst also acknowledging the young person's need to become an independent adult.

Advances in personalization through patient stratification and prediction of clinical outcomes are likely to come from large‐scale real‐world data, including genomic and deeply phenotyped clinical information, together with comorbidity and treatment outcome data ³⁹⁸ . In FEDs, the newly established Eating Disorders Clinical Research Network (ED‐CRN) (www.kcl.ac.uk/research/eating‐disorders‐clinical‐research‐network) aims to capture such data across the full range of ED patients presenting to services across the UK. In addition, studies are needed that integrate information from real‐time deeply phenotyped longitudinal data, using remote measurement technology with active and passive sensing, to provide better delineation of biological and psychological components of different illness and recovery trajectories across FEDs. A number of studies on this topic are in progress, with the ultimately aim of using characteristic data signatures to aid improved illness staging and the development of protocols for just‐in‐time interventions ²⁸ , ³⁹⁹ , ⁴⁰⁰ , ⁴⁰¹ .

An emerging novel precision approach to personalizing psychological treatments, which is not for particular subgroups or clusters of patients, but truly individualized, is via the use of idiographic (one person) network models of ecological momentary assessment symptom data to individually match participants to evidence‐based modules of treatment ²⁹ , ⁴⁰² . In an open case series, this approach was found to be highly acceptable and feasible. A clinical trial is now in progress to assess whether this approach improves outcomes compared to standard psychological treatments for FEDs ⁴⁰³ .

Beyond the realms of personalized psychotherapies, novel brain‐targeted treatments (e.g., brain stimulation approaches) are also emerging as safe and acceptable precision treatments for FEDs ⁴⁰⁴ . These tools allow neuroimaging‐guided targeting of specific brain circuits underpinning specific symptoms, such as low mood, anxiety or emotion regulation, with additional methods of personalizing this approach via individuals’ brain connectivity or EEG patterns or other physiological parameters ⁴⁰⁵ .

Advances in precision nutrition‐based interventions for obesity are emerging ⁴⁰⁶ , with clear relevance to FEDs. Such precision nutrition takes into account both individual‐level and environmental characteristics, such as genetic predisposition, circadian rhythms, physical activity and sedentary behavior, metabolomics, the gut microbiome, and behavioral and socio‐economic characteristics ⁴⁰⁶ . Data for the application of such an approach to FEDs are as yet lacking.

In light of these more recent developments, one challenge for the field is that the most widely used evidence‐based treatments – CBT‐ED and FBT – are based on manuals that are more than 20 years old. These (and other) evidence‐based treatments need to be adapted in light of new insights into comorbidities and major sociocultural shifts, including consideration of under‐served groups, in order to personalize treatment and improve clinical practice ⁴⁰⁷ .

One key recommendation ⁴⁰⁷ is providing patients (and families) with up‐to‐date personalized psychoeducation. Examples of this include the psychoeducational resources for young people with early‐stage EDs on the FREED website, which highlight the malleability of brain and other biological changes in the early illness stages (www.FREEDfromED.co.uk). Such malleability narratives have been found to increase patients’ hope and recovery expectations ⁴⁰⁸ .

Ultimately, the holy grail is the development of precision approaches to personalization of ED care. To drive this field forward, several factors need to be considered. Even if one accepts the utility of measurement‐based precision approaches to personalizing psychological treatments, their implementation is complex. A generic framework for implementing these approaches has been established, giving consideration to evidence supporting them, as well as facilitators and barriers. Demonstrating the value of such approaches to patients and clinicians and getting their “buy‐in” is crucial ²² .

Moreover, the current disparity in research funding (for example, per‐person research funding for schizophrenia in the US and Australia is 69 and 84 times larger than for EDs, even though these disorders are comparable in terms of YPLL ⁸ ) needs to be remedied urgently, to drive the field forward and obtain the improved understanding of underlying neurobiological, genetic, environmental, nutritional and psychosocial drivers of FEDs that is needed for optimal personalization of prevention, treatment and care.

Finally, to achieve truly personalized care, people with lived experience of FEDs and their carers need to be involved in co‐production of new research, treatment and service initiatives designed to address this issue ⁴⁰⁹ , from their inception and throughout.