Progression of Anorexia Nervosa: Longitudinal Staging Framework
By Joanna Steinglass, MD & Enzo Fantin-Yusta, BS
Anorexia nervosa (AN) is a complex and devastating illness. The lifetime prevalence of AN in women is estimated at about 1 percent. The crude mortality rate is among the highest of any psychiatric illness, estimated at 5 percent per decade of illness, and approximately six times that expected among young women, with death most commonly resulting from medical complications of starvation, or from suicide. In addition to mortality costs, AN carries a high global burden of disease owing to disability, the need for medical and psychiatric hospitalizations, high relapse rates, and, often, the need for chronic treatment. Important advances in research have led to increased understanding of some biological and neural mechanisms of illness. And yet, treatment outcomes unfortunately have not improved significantly in recent decades. As interest in personalized medicine increases, the eating disorders field has shown renewed interest in improving understanding of the course of illness in AN.
Acute treatment of AN focuses on renourishment (weight restoration) and often includes inpatient, residential, or day treatment programs (or, for adolescents, family-based therapy). Adolescents commonly have a better prognosis in terms of recovery than adults do, and if it persists, the illness can be difficult to treat. Ongoing outpatient treatment is recommended after weight restoration. Medications have not generally been found to be helpful, though a recent large randomized, controlled trial indicated that olanzapine can help with weight gain among outpatients with AN. Presentation of illness ranges from mild to severe, and course of illness ranges from short-term to chronic.
An illness staging framework may help advance both clinical care and scientific understanding of AN. Staging models aim to characterize illness, with the intent to refine treatment recommendations to fit the stage of illness. The prototypical example of staging models in medicine comes from cancer, where staging is useful in guiding treatment. Some models define discrete stages based on symptom severity, while other models define stages based on an assumption of longitudinal progression. In psychiatric illness, longitudinal progression may mean that the brain changes over time. These changes, in turn, can affect symptom severity. In AN, neuroprogression may relate to (or result from) the effects of starvation.
A longitudinal staging framework for AN has been proposed by Treasure and colleagues (2015). This framework considers both risk factors and neurobiological progression of illness. Recently, our group convened a panel of experts and, through an online, iterative process, achieved consensus about the value of using a longitudinal framework (largely overlapping with the Treasure et al. model). The panel agreed that a longitudinal model of AN would usefully include a subsyndromal phase that can progress to AN, which can either remit or become persistent—and that after remission, relapse can occur (see Box A and Figure 1). Here, we describe these stages and discuss the potential clinical and research utility. Empirical data are needed to substantiate the reliability and validity of these phases.
An eating disorder is characterized, in large part, by maladaptive eating behavior. These symptoms may begin with disordered eating or distorted cognitions, or a combination. Restrictive eating is defined as caloric intake that is less than an individual’s energy requirements and is a common behavior (e.g., dieting). Yet weight loss, which is a defining component of AN, can occur only if restrictive eating behavior is recurrent. As such, for many individuals, there may be a phase of subsyndromal cognitions and behaviors that precedes the underweight, acute illness—by which time restrictive eating has become maladaptive.
In the proposed framework, subsyndromal AN is defined by the presence of restrictive eating, though the individual is not significantly underweight. Descriptively, restrictive eating can include limits on amount or type of food. Excessive or compensatory exercise may also occur, though energy balance is sufficient to maintain weight. Cognitive symptoms in subsyndromal AN (as in AN) can include a disturbance in the perception and experience of one’s body, a desire to lose weight despite being of normal or low weight, intense fears of weight gain, and obsessions with thinness and/or dieting.
Identifying subsyndromal AN may be challenging. If binge eating, purging, or laxative use is present, it may be difficult to distinguish subsyndromal AN from a subsyndromal phase of a different eating disorder. In addition, it may be difficult to distinguish subsyndromal pathological cognitions from those of a normal dieter. Yet many patients with AN are able to describe a subsyndromal phase of their illness, in which they experienced maladaptive cognitions and behaviors prior to meeting criteria for AN. Epidemiological studies indicate high rates of eating disorders that have not required treatment, suggesting that for some individuals, subsyndromal AN may resolve without progressing to full AN. It is tempting to consider that identification of subsyndromal AN, and delivery of targeted interventions, might serve to prevent the onset of severe illness.
Full Syndrome AN
The diagnostic criteria for AN include a combination of being significantly underweight, having a fear of weight gain (inferred or expressed), and possessing an overvaluation or distortion of body image. Underweight is defined as weight less than what is minimally normal for the individual’s age, height, sex, developmental trajectory, and physical health. For adults, this is commonly determined by body mass index (BMI) and, among children and adolescents, as BMI percentile. The longitudinal framework, and the possibility of neuroprogression, highlights that it may be useful to consider stages, or time frame qualifiers, within the course of AN. However, in the absence of empirical data, this is challenging, as there are no definitive answers for determining the appropriate hinge points between these stages.
One challenge is how or whether to distinguish an early phase of illness. There are data showing that intervening earlier in the course of AN leads to better outcomes (greater chance of remission and recovery). However, most of these data come from studies of adolescents. Among adults, longitudinal research often (but not always) shows that longer duration of illness is associated with worse outcomes. The panel of experts did not fully agree about whether an “early” stage qualifier should be considered a discrete entity. At the same time, there was agreement that if there were such a stage, the appropriate time frame to consider would be the first year of illness. To date, there are insufficient data to clarify a specific “early stage AN” time frame.
A similar challenge exists for qualifying AN as “persistent” (also termed chronic, or severe and enduring). While treatment is successful with approximately half of adult patients, at times AN persists despite multiple courses of treatment. In addition, some patients have been ill for many years before presenting for treatment. The expert panel had some agreement that persistent AN is defined by the presence of illness for more than three years (77 percent agreement, considered “near-consensus”). Others have proposed that the illness be considered persistent if the patient has engaged in at least two evidence-based treatments without success. A challenge in the determination of persistent illness is that remission can occur even after long-standing illness. Thus, while it may be that successful interventions for persistent illness differ from those earlier in the course of illness, the specific differences in treatment recommendations are not yet known.
Clinical Implications of a Longitudinal Staging Framework
Identification of a subsyndromal AN stage is most useful if it allows for prevention of illness. Some programs have been created to prevent eating disorders more broadly, and they seem to have some utility. However, the data are stronger for prevention of pathological binge eating than for restrictive eating and AN. Notably, such interventions (e.g., The Body Project) have had a larger effect among individuals with body dissatisfaction, internalization of the cultural thin-ideal standard of (female) beauty, negative affect, and who are engaging in dieting—potential risk factors for development of an eating disorder. Analogously, in schizophrenia, identification of a prodromal phase has yielded successful preventive interventions (e.g., use of long-chain omega-3 fatty acids to prevent the development of psychosis). Thus, the possibility of improving outcomes through prevention is an appealing feature of a longitudinal framework.
Similarly, there are data to suggest that early intervention in AN is more successful. Among adolescents with AN, family-based treatment (FBT) is associated with good outcomes in up to 70 percent of patients. This treatment focuses on empowering parents to manage the patient’s daily meals and interrupt compensatory behaviors. Gradually, the patient regains independence as symptoms recede. This approach requires fewer health care resources than some of the more-intensive settings that are most utilized by adults. Among adults, acute treatment for AN includes behaviorally based treatment in a structured setting (i.e., inpatient, residential, or day hospital), generally followed by outpatient care. It may be that when illness is identified early, less-intensive health care is required.
Once the illness becomes entrenched, treatment becomes more difficult. New approaches suggest thinking about habit formation in illness progression, and interruption of maladaptive habits in treatment models. Some have argued in favor of shifting toward a harm reduction model once AN is considered persistent. In the absence of data for when this is necessary and without a uniform definition of persistent AN, the decision to shift treatment goals may not be warranted.
Research Implications of a Longitudinal Staging Framework
Improvements in treatment outcomes often come through the understanding of pathophysiology. Biomarkers, or biological indicators of illness, are often sought after to improve the precision of treatment. In AN, a genome-wide association study provides some preliminary evidence that there may be biomarkers worth studying. To date, however, there are no biomarkers to distinguish the stages of the longitudinal framework for AN. It would be useful to have clear biological indicators of starvation—a uniform definition of underweight—yet medical presentations of AN vary tremendously by individual.
Neural mechanisms underlying AN are beginning to be established. For example, a few studies have converging data that suggest reward system abnormalities. A longitudinal staging framework, in which hypotheses can be tested among a more homogeneous population of patients within a given illness stage, may yield more information about mechanisms of illness and the possibility of neuroprogression. For example, one manifestation of neuroprogression in schizophrenia is enlargement of the brain’s lateral ventricles, a change that is less profound in earlier stages of the illness. There is converging evidence in AN that the brain circuitry involved in reward processing and in habit formation may be aberrant. Understanding whether these changes progress with the effects of starvation, or starvation over time, may be useful in identifying stages of illness. Longitudinal brain imaging studies of cohorts of patients with AN could shine light on the utility of imaging and neuroprogression in the staging of AN.
A longitudinal framework for AN considers progression from subsyndromal stage to early illness to AN, and then to persistent AN, with opportunities for remission at any stage. The trajectory of illness is presumably influenced by genetics and biological factors, social and cultural context, timing of the identification of the illness, and engagement in appropriate treatment modalities.
Although some aspects of each stage may appear arbitrary, a longitudinal staging model has much potential value, if validated. One example comes from schizophrenia, where the validation of a longitudinal staging model led to the identification of a prodromal phase, in which patients exhibit some of the negative symptoms of schizophrenia and may have “near-psychotic” experiences. The prodrome can last months to years, and it is usually identified retrospectively after the individual has progressed into schizophrenia. Some studies have found interventions aimed at the prodrome to be efficacious in preventing development of schizophrenia. Similar principles may be applicable to the prevention of subsyndromal AN progressing to full AN. From a research perspective, consensus within this model will allow for more standardized methods across groups to improve collaborations and comparisons. This may yield identification of biological markers to advance understanding of the mechanisms of the illness. The longitudinal staging framework may lead to identification of stage-appropriate interventions and may allow for improved reproducibility of findings across research. Advances in clinical care and research will improve understanding of this difficult illness for patients and their families.
Box A: Case Illustration of Stages of Anorexia Nervosa
Patient ZZ is a 16-year-old attending sleepaway camp for the summer. She is enjoying herself and making new friends, but she does not like the food. ZZ begins to eat selectively, in smaller portions, and refuses some foods. When ZZ returns home, her restrictive eating continues. She used to love the family tradition of Steak Fridays, but now refuses meat and eats spinach instead. She begins to talk about her body regularly. Soon, her parents note that she has started counting how many calories she consumes in a day.
Path 1: Subsyndromal and Resolution: Parents notice this is a change in her behavior and are perplexed by it. Her weight is appropriate for her age and height. The question becomes: To what extent is this a problem? The changes noted by her parents and the negative body image may help distinguish dieting from subsyndromal AN.
ZZ and her mother bring up these changes to her pediatrician, who recognizes this as potentially concerning. Clinical attention at this stage may focus on adequate nutrition, including dietary fat, and positive body image. ZZ engages with groups (e.g., The Body Project) that help her critique her culture’s thin-body ideal, reduce body dissatisfaction, and establish a more fulfilling relationship with food and physical activity.
Path 2: Progression to AN: Over the course of the next two to three years, ZZ continues her low-fat intake and restrictive eating, and increases her exercise (does squats and jumping jacks prior to meals), and her body image worsens. She declines eating meals with her friends, she has difficulty concentrating in class, and she stays up late to complete her assignments. Her menstrual cycles stop. She goes to her primary care doctor, where her BMI is calculated as 16.8 kg/m2 (normal BMI is between 18.5 kg/m2 and 25 kg/m2). The doctor identifies this as anorexia nervosa and recommends treatment.
2a) AN vs. Early AN: ZZ’s illness progresses over the course of three years after summer camp. Her BMI has been in the underweight range for an undetermined amount of time. It is likely that it has been more than a year, such that the qualifier of early AN is no longer applicable. Unfortunately, there are no biological tests to distinguish early AN. ZZ follows her doctor’s recommendation for treatment.
- b) Treatment: Although it is a terrifying prospect for ZZ, she attends an inpatient weight restoration program. She appreciates the group therapy sessions and speaking with other people who are going through a similar experience. She works with a psychiatrist, a social worker, and a nutritionist. ZZ achieves full weight restoration and is discharged to outpatient care, where she is able to maintain healthy eating, a healthy weight, and a sustained remission from illness.
- c) Persistent AN: Despite initial success in inpatient treatment, ZZ finds that she loses weight whenever she leaves the structured treatment setting. She continues to struggle with low weight and restrictive eating, and her AN persists for several more years after she graduates from college. Though somewhat discouraged by lack of recovery after multiple attempts at treatment, she tries again. With another round of weight restoration, and renewed interest in her life goals, she has new insights about the behavioral routines that had been undermining her success. This time, she maintains her health and the remission of her symptoms.
Figure 1: Longitudinal Progression of Anorexia Nervosa—A Case Example (see Box A)
The figure depicts the stages of AN and how a patient might progress through those stages (blue), or might remit from AN with treatment (white). Treatment (orange) at each stage may interrupt the illness course.
About the authors:
Dr. Joanna Steinglass is a Professor of Psychiatry and the Associate Director of the Center for Eating Disorders at Columbia University Medical Center and the New York State Psychiatric Institute. Dr. Steinglass graduated from Amherst College and received her medical degree from Harvard Medical School.
She completed her psychiatry training and research fellowship in eating disorders at Columbia and has been on faculty ever since. Her research aims to understand the mechanisms of anorexia nervosa and to develop mechanism-based treatments.
Through the research clinic, patients who are interested in and eligible for research participation are able to receive treatment at no-cost. To learn more about research participation see link here: https://www.columbiapsychiatry.org/research-clinics/eating-disorders-clinic or at the feedblog.com. To reach the clinic, you can call 646-774-8066 or email email@example.com.
Enzo Fantin-Yusta is an MD candidate at Columbia University Vagelos College of Physicians & Surgeons, class of 2021. He earned a Bachelor of Science in Neuroscience from the University of Michigan. Mr. Fantin-Yusta will be pursuing residency training and a career in Psychiatry.
Association AP. Treatment of patients with eating disorders, third edition. American Journal of Psychiatry. 2006;163:4-54.
Attia E, Steinglass JE, Walsh BT, Wang Y, Wu P, Schreyer C, Wildes J, Yilmaz Z, Guarda AS, Kaplan AS, Marcus MD. Olanzapine versus placebo in adult outpatients with anorexia nervosa: A randomized clinical trial. The American Journal of Psychiatry. 2019;1[76:44]9-456.
Brockmeyer T, Friederich HC, Schmidt U. Advances in the treatment of anorexia nervosa: A review of established and emerging interventions. Psychological Medicine. 2017:1-37.
McGorry P, Keshavan M, Goldstone S, Amminger P, Allott K, Berk M, Lavoie S, Pantelis C, Yung A, Wood S, Hickie I. Biomarkers and clinical staging in psychiatry. World Psychiatry. 2014;[13:21]1-223.
Steinglass JE, Glasofer DR, Dalack M, Attia E. Between wellness, relapse, and remission: Stages of illness in anorexia nervosa. International Journal of Eating Disorders. 2020;[53:10]88-1096.
Steinglass JE, Walsh BT. Neurobiological model of the persistence of anorexia nervosa. Journal of Eating Disorders. 2016;4.
Treasure J, Stein D, Maguire S. Has the time come for a staging model to map the course of eating disorders from high risk to severe enduring illness? An examination of the evidence. Early Intervention in Psychiatry. 2015;[9:17]3-184.