Abstract
Background
Nutritional rehabilitation is widely considered a prerequisite for recovery in anorexia nervosa (AN), given the profound effects of malnutrition on cognitive, emotional, and behavioral functioning. However, whether psychiatric improvement during residential treatment is directly coupled with the degree of nutritional restoration remains insufficiently understood. This study investigated longitudinal changes in body composition and psychosocial outcomes during residential psychiatric rehabilitation for AN, with the aim of informing the evaluation and planning of intensive rehabilitation programmes.
Methods
This exploratory longitudinal observational study included 19 female patients with AN or atypical AN (mean age 20.4 ± 5.6 years) admitted to a specialized residential programme. Multi-frequency bioelectrical impedance analysis (BIA) was performed at admission (T0) and discharge (T1; ≈ 5 months). Diagnoses were confirmed via the Eating Disorder Examination (EDE) and psychiatric comorbidities via the SCID-5. Psychopathology and body image were assessed using the EDE-Q, EDI-3, SCL-90, CIA, EDS-R, and BUT. Wilcoxon signed-rank tests evaluated T0–T1 changes; Spearman correlations assessed nutritional-psychiatric associations. Benjamini-Hochberg correction was applied to all secondary outcomes.
Results
BMI increased significantly (p = 0.042, r = 0.35), whereas FFM and SMM showed non-significant medium-effect-size upward trends. Large and significant improvements were observed in eating disorder psychopathology (EDE-Q: r = 0.74), psychological distress (SCL-90 GSI: r = 0.66), functional impairment (CIA: r = 0.62), exercise dependence (EDS-R: r = 0.72), and body image disturbance (BUT GSI: r = 0.88, the largest effect size in the study). Sixteen of 18 EDI-3 subscales, all nine SCL-90 dimensions, all seven EDS-R subscales, and all seven BUT-A indices improved significantly after false discovery rate correction. Nutritional and psychiatric change scores were not significantly correlated across any pairing.
Conclusions
Five months of residential treatment produced large-magnitude psychiatric improvements alongside modest nutritional gains. The absence of significant somatic-psychiatric correlations should be interpreted with caution: the study had only 37.5% power to detect medium-sized associations (rs = 0.40), yielding a Type II error probability of approximately 62.5%. These findings are hypothesis-generating and do not permit inference about the degree of coupling between nutritional and psychiatric recovery in AN.