Frontiers in Oncology · Nov 10, 2026

The association between sleep disturbance and nutritional status in cancer chemotherapy patients: an exploratory study on the mediating role of frailty

By Xiaoyan Hao, Ling Zhou, QIN CHEN et al.

Background: Chemotherapy frequently causes concurrent sleep disorders and malnutrition in cancer patients, but the mediating role of frailty remains underexplored. Previous cross-sectional studies are highly susceptible to biases from overlapping measurement items (e.g., weight-loss criteria), age collinearity, and unadjusted confounders. This exploratory study aimed to examine whether frailty mediates the relationship between sleep disturbance and malnutrition, rather than to verify causality. Methods: We conducted a cross-sectional survey of 125 hospitalized chemotherapy patients using the modified Patient-Generated Subjective Global Assessment (mPG-SGA), Fried Frailty Phenotype (FP), and Pittsburgh Sleep Quality Index (PSQI). Mediation analysis used 5,000 nonparametric percentile bootstrap resamples, adjusting for age, sex, cancer stage, and tumor site. To address measurement and confounding biases, sensitivity analyses included removing weight-loss items, removing the age item from mPG-SGA, reverse models, and age subgroup analyses. An exploratory age × PSQI interaction tested moderation. Results: More severe sleep disturbance was associated with worse malnutrition (ρ = 0.305, P < 0.001) and frailty (ρ = 0.200, P = 0.025). The unadjusted indirect effect was significant (B = 0.071, 95% CI: 0.012 to 0.145) and remained so after partial adjustment for age, sex, and TNM stage (B = 0.061, 95% CI: 0.006 to 0.134), but became non-significant after full adjustment including tumor site (B = 0.057, 95% CI: -0.001 to 0.135). Sensitivity analyses removing weight-loss items (B = 0.046, 95% CI: -0.001 to 0.110) and addressing age collinearity (B = 0.056, 95% CI: -0.001 to 0.133) were also non-significant. Neither age subgroup showed a significant indirect effect, and the age×PSQI interaction was not statistically significant (P = 0.266).

The paperJournal sitePopulations

AgingResearch.News