Sleep-Related and Inflammation-Related Indirect Pathways in the Association Between Eczema and Depression: A Cross-Sectional Analysis of the National Health and Nutrition Examination Survey (NHANES) Database

Background: Eczema/atopic dermatitis (AD) is associated with depression, but the potential indirect pathways involving sleep disorders and peripheral inflammation remain uncertain. We examined sleep-related and inflammation-related indirect effects in the eczema/AD-depression association. Methods: We analyzed cross-sectional data from 2,071 adults aged 20-59 years who participated in the 2005-2006 National Health and Nutrition Examination Survey. Depression was defined as a Patient Health Questionnaire-9 score ≥10. Physician-diagnosed sleep disorder, the systemic immune-inflammation index (SII), and the aggregate index of systemic inflammation (AISI) were evaluated as potential indirect pathways. Survey-weighted linear probability models and 5,000 design-based Rao-Wu bootstrap replicates were used to estimate indirect effects on the risk-difference (RD) scale. Inflammation-related effects were additionally assessed using equivalence testing. Results: Eczema/AD was associated with a higher prevalence of depression (total-effect RD = 0.0689, 95% bootstrap confidence interval (CI) 0.0233-0.1142; two-sided P = 0.002). The sleep-related indirect estimate was positive but imprecise (RD = 0.00860, 95% bootstrap CI -0.00024 to 0.01888; two-sided P = 0.058). Indirect effects involving SII or AISI were negligible, and their 90% CIs were contained within the prespecified equivalence bounds of ±0.00689. Similar findings were obtained for the neutrophil-to-lymphocyte ratio, eosinophil count, total immunoglobulin E, and high-sensitivity C-reactive protein. Conclusions: The findings are consistent with a possible sleep-related indirect pathway between eczema/AD and depression, although the estimate did not meet the conventional two-sided significance threshold. Indirect effects involving the measured peripheral inflammatory markers were negligible relative to the prespecified equivalence margin. Because of the cross-sectional design, temporal ordering and causal mediation cannot be established.