Serum osmolality as a potential marker of inflammatory bowel disease in patients with anxiety: a cross-sectional analysis based on the MIMIC-IV database

Inflammatory bowel disease (IBD) is common among individuals with anxiety and substantially impairs quality of life. Serum osmolality (SOSM), a key biomarker of metabolic and fluid homeostasis, has been associated with various chronic inflammatory and immune-mediated disorders. However, clinical evidence regarding the SOSM-IBD association in anxious populations remains limited. This cross-sectional study aimed to examine whether SOSM is independently associated with IBD prevalence among hospitalized patients with anxiety in the MIMIC-IV database. We extracted data on inpatients with anxiety admitted to general wards from the MIMIC-IV database. We used nested 10-fold cross-validation to assess the discriminative ability of baseline SOSM. Univariate analysis and variance inflation factor (VIF) tests were used for variable screening and multicollinearity evaluation. We applied stabilized inverse probability weighting (sIPTW) to mitigate baseline confounding. Multivariable logistic regression and Firth penalized logistic regression were used to examine the cross-sectional association between SOSM and IBD, and restricted cubic spline (RCS) analysis was used to characterize nonlinear dose-response relationships. XGBoost combined with SHapley Additive exPlanations (SHAP) was used to quantify feature importance. We assessed robustness through sensitivity analyses (unweighted regression, inverse probability of selection weighting, narrow definition of anxiety, and E-value estimation) and subgroup analyses. We included 9,511 participants with anxiety (211 with IBD and 9,300 without). Nested 10-fold cross-validation showed moderate discriminative ability of SOSM for IBD (AUC, 0.695). After adjustment for confounders using sIPTW, high SOSM (≥ 296.71 mOsm/kg) was independently associated with IBD prevalence (odds ratio [OR], 14.36; 95% CI, 9.77–21.12; P < 0.001). RCS analysis revealed a significant nonlinear dose-response relationship (P for nonlinearity < 0.001). XGBoost-SHAP analysis ranked SOSM as the second most important feature after platelet count. Subgroup analyses showed consistent associations across demographic and clinical strata. Multiple sensitivity analyses further supported the robustness of this association. In this cross-sectional cohort of hospitalized patients with anxiety, high SOSM was independently and positively associated with IBD prevalence. Because of the cross-sectional design, causal inferences cannot be drawn. Nevertheless, SOSM may serve as a low-cost, adjunctive quantitative marker for IBD in this population, pending validation in prospective studies.