Abstract
This study used cluster analysis to explore clinically relevant subgroups of adult patients with anorexia nervosa (AN). Patients were clustered based on their body mass index (BMI), eating disorder symptomatology, anxiety and depression symptoms and autistic characteristics. The difference between clusters in work and social functioning, duration of illness, bingeing and purging behaviour, previous hospitalisations and number of comorbidities was also investigated. Two meaningful clusters emerged: a higher symptoms cluster with more severe eating pathology, anxiety, depression, and more autistic traits, and a second cluster with lower symptoms. BMI did not make major contributions to cluster formation. The higher symptoms cluster also reported lower self-efficacy to change, more previous hospitalisations, comorbid diagnoses, binge eating and purging behaviours and use of psychotropic medication. Our findings suggest that weight alone may not be a significant severity indicator amongst inpatients with AN, and targeted treatment of AN should consider a broader range of symptom severity indicators.