Longitudinal Changes in Insomnia, Sleep Continuity, and Sleep Timing During Residential Treatment for Eating Disorders: A Six-Month Observational Study

Abstract

Background: Sleep disturbances are a clinically relevant and transdiagnostic dimension of eating disorders (EDs), particularly during adolescence and young adulthood. Although early improvements in insomnia severity during residential treatment have been described, the longitudinal trajectory of sleep disturbances, habitual sleep timing, and eating disorder psychopathology across a six-month residential stay has not been systematically characterised. Methods: This prospective real-world observational study enrolled 30 female adolescents and young adults (aged 14–22 years) in a small exploratory cohort, predominantly with anorexia nervosa (AN; 83.3%) and a minority with bulimia nervosa (BN; 16.7%), all with clinically significant insomnia at admission (Insomnia Severity Index [ISI] > 14). All sleep outcomes were based on subjective self-report; no objective sleep assessment (actigraphy or polysomnography) was available. Participants were consecutively recruited from the Regional Residential Centre for Eating Disorders “Mariconda”, ASL Salerno, Italy. Of 38 consecutive admissions assessed for eligibility, 8 were excluded (ISI ≤ 14 at admission: n = 5; incomplete baseline data: n = 3), yielding a final enrolled sample of n = 30. All 30 participants completed the six-month residential programme; no drop-outs occurred. ISI was assessed monthly (T0–T6); PSQI and EDE-Q were assessed at T0, T3, and T6; diary-derived sleep timing (sleep midpoint) was assessed monthly (T0–T6), while diary-derived sleep continuity parameters (SOL, WASO, TST, SE%) were available at T0, T1, T3, and T6. Linear mixed-effects models (LMM) and bootstrap-resampled paired analyses were used. Results: ISI declined progressively from 21.1 ± 2.5 at baseline to 7.4 ± 2.7 at month 6 (Δ = −13.7; 95% CI [−14.4, −13.1]; p < 0.001). By month 6, 56.7% of participants had achieved remission (ISI < 8) and 100% had met response criteria (ΔISI ≤ −8). Sleep midpoint advanced progressively by 86 min over six months (from ~04:41 to ~03:16 after midnight), with no significant association with BMI or pharmacological treatment in linear mixed-effects models. EDE-Q global scores decreased from 4.57 ± 0.49 to 2.06 ± 0.58. Changes in ISI were not correlated with changes in EDE-Q at month 3 (r = 0.042, p = 0.826) but were significantly correlated at month 6 (r = 0.480, p = 0.007). All pharmacological regimens remained stable throughout the six-month observation period. The association between improvement in insomnia severity and eating-disorder psychopathology was not evident at month 3 but emerged at month 6, suggesting that these domains may change at different rates during residential treatment. Given the exploratory nature of the analyses, p-values are reported without adjustment for multiple comparisons and should be considered hypothesis-generating. Conclusions: Across six months of residential treatment, progressive improvements were observed in insomnia severity, subjective sleep continuity, habitual sleep timing, BMI, and eating-disorder psychopathology. In the absence of a control group, spontaneous remission, regression to the mean, and non-specific components of care, including the routinely administered Passiflora incarnata L. adjunct, cannot be individually distinguished as contributors to the observed changes, and these findings should be interpreted within the constraints of an uncontrolled observational design relying on subjective sleep measures. Controlled studies are needed to determine whether these changes are directly attributable to treatment or primarily reflect concurrent weight restoration and clinical improvement.