Abstract
Recent studies have shown a significant association between iron and the development and prognosis of sepsis, but the relationship between iron levels and mortality in sepsis patients remains unclear, with previous studies examining this relationship under a linear assumption. This retrospective observational study aimed to assess the possible non-linear relationship between serum iron (SI) levels and 28-day all-cause mortality (28-DACM) in individuals with sepsis. We used multiple imputation for data with less than 30% missing values and developed Cox models to calculate the hazard ratio (HR) with a 95% confidence interval (CI) for the main outcome (28-DACM). To accurately assess the relationship between SI levels and the HR for 28-DACM, we utilized a restricted cubic spline (RCS) regression model with five knots. We also conducted subgroup analyses to evaluate the robustness of the primary results. The study found that SI levels upon ICU admission are an independent predictor of 28-DACM in sepsis patients, showing a J-shaped correlation with mortality. Patients with extremely high and low SI levels had higher mortality compared to those within the normal range. However, this study has limitations inherent to its retrospective observational design, such as potential bias and unmeasured confounding factors. Future studies should address these limitations through prospective designs.