Mary was 16 when she was first diagnosed with anorexia nervosa. Her parents remember the hospital visits, the calorie logs, and the quiet panic that came with every skipped meal. Over time, her symptoms shifted. Some days she binged and purged; other days she seemed to be improving. Her doctors debated revising her diagnosis to better align with her evolving symptoms and support needs. Her family just wanted to know: was she getting better and would she be ok? Although Mary’s story is fictional, it reflects the lived experience of many people with eating disorders and illustrates the complex questions we continue to ask about diagnosis, recovery and genetic risk.
Eating disorders (EDs) such as anorexia nervosa (AN), bulimia nervosa (BN) and eating disorder not otherwise specified (EDNOS) affect millions worldwide, contributing to an estimated 3.3 million healthy life-years lost annually (van Hoeken & Hoek, 2020). Yet despite their severity, the course of these illnesses is rarely linear. Some individuals transition between diagnoses, while others appear to recover, at least temporarily. Previous studies have reported wide-ranging estimates for diagnostic transitions and remission, often based on small or clinical samples (Eddy et al., 2008; Tozzi et al., 2005). What drives these changes? Could biological factors, such as genetics, play a role?
A new study from Denmark (Abdulkadir et al., 2025) uses national health registers and genetic data from over 10,000 individuals to explore two key questions: (1) how often do people with EDs change diagnoses or recover, and (2) can polygenic scores (PGSs), which reflect genetic predisposition to complex traits, help predict these outcomes?
Eating disorders often follow complex paths. This study explores how genetics may help explain them.
Methods
Researchers analysed data from 10,565 individuals born in Denmark between 1981 and 2009 who had been diagnosed with AN, BN, or EDNOS. All individuals had at least two ED-related hospital contacts recorded in the Danish national health registers between 1995 and 2018.
The study examined three outcomes describing the course of illness:
- Number of ED episodes = total number of distinct treatment episodes reconstructed from ED-related care; a new episode begins after a treatment gap or a change in diagnosis.
- Time to diagnostic transition = length of time between an individual’s change from one ED diagnosis to another.
- Time to presumed remission = a two-year period without any ED-related hospital contact.
Genetic data came from dried blood spot samples collected shortly after birth as part of Denmark’s routine neonatal screening programme. These samples were genotyped as part of national psychiatric genetics projects (iPSYCH, ANGI-DK, and EDGI-DK) and linked to individual clinical records.
Using these genetic data, the researchers calculated 422 polygenic scores (PGSs) capturing inherited predisposition across a wide range of domains, including psychiatric traits (e.g., depression, anxiety, ADHD), metabolic markers (e.g., BMI, insulin resistance), and behavioural tendencies (e.g., risk-taking, physical activity), and broader indicators of well-being (e.g., life satisfaction, health rating).
To explore genetic influences, associations between each PGS and the three outcomes were tested using Cox proportional hazards models, adjusting for ancestry and multiple comparisons.
Results
The sample was predominantly female (92.6%), with European ancestry (85.5%). At the time of their first diagnosis, 65.3% were diagnosed with AN, 13.4% with BN, and 21.3% with EDNOS. The average age at initial diagnosis was 18 years, and participants were followed for a mean duration of 8.9 years.
Diagnostic transitions
Shifts from one ED diagnosis to another were relatively uncommon. Among those initially diagnosed with AN, 5.4% later transitioned to BN and 9.7% to EDNOS. For BN, 11.6% transitioned to AN and 12.3% to EDNOS. EDNOS showed the highest overall transition rate (23.1%), most frequently to AN (17.8%). These transitions typically occurred early in the illness course, usually within five years after diagnosis (mean time = 4.7 years). This suggests that diagnostic shifts, when they happen, tend to occur early in the illness course.
Presumed remission
A two-year period without any ED-related hospital contact, used as an indicator of presumed remission, was observed in the majority of participants. Specifically, 86.9% of individuals with AN, 89.8% with BN, and 89.0% with EDNOS experienced remission.
On average, remission occurred approximately 1.6 years after the initial episode and lasted an average of 5.3 years. However, a relapse following remission was observed in 15% of the sample, with the highest rate among individuals with BN (20.5%).
Genetic influences on illness course
The study also examined whether genetic liability for other psychiatric and health-related traits helped explain who transitions versus who recovers. A higher genetic risk for major depressive disorder and multi-site chronic pain was associated with a 15% greater likelihood of transitioning from AN to either BN or EDNOS, suggesting that co-occurring affective or somatic vulnerability may make diagnostic change more likely.
In contrast, remission from AN was more common among individuals with a higher genetic predisposition to traits related to metabolism and socioeconomic context. Specifically, higher PGSs for leg fat percentage and financial difficulties were associated with an 8% and 5% greater likelihood of remission, respectively. These patterns suggest that both metabolic factors and life circumstances may facilitate recovery.
For EDNOS, higher genetic liability to mood swings was linked to a 10% greater likelihood of remission, possibly because mood variability increases the likelihood of clinical monitoring or support. On the other hand, individuals with a higher genetic predisposition toward a positive overall health rating were 12% less likely to enter remission, which may reflect lower treatment-seeking among those who perceive themselves as healthy despite ongoing symptoms.
Recovery from eating disorders isn’t rare: nearly 9 in 10 people with eating disorders reached remission, and genetic predispositions influence this. However, relapse following remission was also observed in many individuals.
Conclusions
This large registry-based study found that most individuals with EDs don’t switch diagnoses but many do experience sustained periods without hospital contact, interpreted as presumed remission. Nearly nine in ten participants met criteria for presumed recovery, often within two years of diagnosis.
The authors also found that genetic predispositions related to psychiatric, somatic, and socioeconomic traits may influence both the likelihood of diagnostic change and the probability of recovery in individuals with EDs. The authors concluded:
Most patients with an ED did not experience diagnostic transitions but were more likely to experience a period of presumed remission. Both diagnostic transitions and presumed remission have a significant polygenic component.
Genetic factors may help explain who with an eating disorder is most likely to recover, relapse, or remain ill; providing insights that could support more personalised treatment and care.
Strengths and limitations
This study is notable for its scale and design. It included data from over 10,000 individuals diagnosed with EDs in Denmark and followed them for nearly nine years. By using national health records, the researchers captured almost all hospital-treated cases of AN, BN, or EDNOS during the study period. The genetic analysis was also extensive, covering over 400 different traits, which gave a wide-angle view of how biology might influence recovery or changes in diagnosis.
However, several limitations affect how confidently we can apply these findings. First, the study only included people who were treated in hospitals. This means it may have missed those who were seen in outpatient clinics, private practices, or who never sought formal treatment. These individuals may have different illness trajectories, so the results may not reflect the full spectrum of ED experiences.
Second, diagnoses were based on registry coding, which depends on how clinicians record cases. This might introduce observer bias (when a researcher’s expectations influence what they see or record). EDNOS, in particular, is a vague category that often reflects diagnostic uncertainty or fluctuating symptoms, making it difficult to interpret transitions between diagnoses with precision.
Third, the definition of remission – two years without hospital contact – is practical but limited. It may reflect people stepping away from services rather than genuinely recovering, and it doesn’t capture whether symptoms have improved, persisted, or worsened. Individuals who continue to struggle but manage their condition outside hospital settings could be misclassified as “recovered.” Although the use of national registers reduces the risk of losing track of participants (i.e., attrition bias), the absence of symptom-level information makes it difficult to interpret what remission truly means in clinical terms.
Fourth and finally, most participants were of European ancestry. Genetic findings may not apply to people from other backgrounds, and future studies should include more diverse populations.
By linking genetic data with national health records, this study offers a detailed picture of recovery and diagnostic stability in eating disorders, showing that long-term remission is common and that genetic factors help explain differences in illness course.
Implications for practice
This study provides a reassuring yet cautiously optimistic message:
Periods without hospital contact are common among people with eating disorders, suggesting that many experience phases of improvement or stability, even within hospital-treated populations. Most individuals remained within their initial diagnosis, and nearly nine in ten had at least one 2-year period without hospitalisation following diagnosis. While this was used as an indicator of presumed remission, it may not capture all aspects of recovery. Some individuals might have continued treatment outside hospital settings, moved away from national coverage, or managed their condition independently.
These findings highlight the importance of early intervention, regular follow-up, and continued support beyond inpatient care. Clinicians should recognise that diagnostic stability is typical, but remission inferred from registry data should be interpreted with care. Where possible, direct clinical follow-up or patient-reported outcomes could help determine whether an absence of hospital contact reflects true recovery or withdrawal from services.
The genetic findings suggest new possibilities for personalised care. For example, people with a higher genetic risk for depression or chronic pain may be more vulnerable to changes in diagnosis and could benefit from closer monitoring. On the other hand, those with genetic predispositions to higher body fat may be more likely to recover from AN, indicating that certain biological traits may help protect against restrictive behaviours. While genetic data is not yet used routinely in clinical settings, this study shows its potential to inform care strategies in the future (Riess et al., 2024).
Recovery is not only possible, but likely. Understanding how genetic factors shape the course of EDs may help clinicians offer more targeted and compassionate support. These findings also have policy relevance. Eating disorders often begin in adolescence, and while remission can happen quickly, relapse is also common (Miskovic-Wheatley et al., 2023). Public health systems should ensure that care does not end with hospital discharge. Long-term follow-up and access to outpatient and community services are essential, especially for those who disengage from formal treatment but remain at risk.
As someone who works with large-scale cohort data, I see the value of registry studies in identifying broad patterns. But I also recognise their limits. A two-year gap in hospital contact may look like recovery in the data, yet it could mean something very different in real life. It might reflect someone managing alone, without support. That gap between data and experience matters. It reminds us that recovery is not just a clinical outcome but a personal journey.
Periods without hospital contact may signal recovery in eating disorders, but true remission requires deeper clinical and personal insight.
Statement of interests
None to declare.
Links
Primary paper
Abdulkadir, M., Larsen, J. T., Clausen, L., Hübel, C., Albiñana, C., Thornton, L. M., Vilhjálmsson, B. J., Bulik, C. M., Yilmaz, Z., & Petersen, L. V. (2025). Descriptives and Genetic Correlates of Eating Disorder Diagnostic Transitions and Presumed Remission in the Danish Registry. Biological Psychiatry, 98(3), 230–236. https://doi.org/10.1016/j.biopsych.2025.01.008
Other references
Eddy, K. T., Dorer, D. J., Franko, D. L., Tahilani, K., Thompson-Brenner, H., & Herzog, D. B. (2008). Diagnostic crossover in anorexia nervosa and bulimia nervosa: Implications for DSM-V. The American Journal of Psychiatry, 165(2), 245–250. https://doi.org/10.1176/appi.ajp.2007.07060951
Miskovic-Wheatley, J., Bryant, E., Ong, S. H., Vatter, S., Le, A., Aouad, P., Barakat, S., Boakes, R., Brennan, L., Bryant, E., Byrne, S., Caldwell, B., Calvert, S., Carroll, B., Castle, D., Caterson, I., Chelius, B., Chiem, L., Clarke, S., … National Eating Disorder Research Consortium. (2023). Eating disorder outcomes: Findings from a rapid review of over a decade of research. Journal of Eating Disorders, 11(1), 85. https://doi.org/10.1186/s40337-023-00801-3
Riess, O., Sturm, M., Menden, B., Liebmann, A., Demidov, G., Witt, D., Casadei, N., Admard, J., Schütz, L., Ossowski, S., Taylor, S., Schaffer, S., Schroeder, C., Dufke, A., & Haack, T. (2024). Genomes in clinical care. NPJ Genomic Medicine, 9, 20. https://doi.org/10.1038/s41525-024-00402-2
Tozzi, F., Thornton, L. M., Klump, K. L., Fichter, M. M., Halmi, K. A., Kaplan, A. S., Strober, M., Woodside, D. B., Crow, S., Mitchell, J., Rotondo, A., Mauri, M., Cassano, G., Keel, P., Plotnicov, K. H., Pollice, C., Lilenfeld, L. R., Berrettini, W. H., Bulik, C. M., & Kaye, W. H. (2005). Symptom fluctuation in eating disorders: Correlates of diagnostic crossover. The American Journal of Psychiatry, 162(4), 732–740. https://doi.org/10.1176/appi.ajp.162.4.732
van Hoeken, D., & Hoek, H. W. (2020). Review of the burden of eating disorders: Mortality, disability, costs, quality of life, and family burden. Current Opinion in Psychiatry, 33(6), 521–527. https://doi.org/10.1097/YCO.0000000000000641
