Early serum albumin dynamics in sepsis differ by albumin source and have implications for nutritional assessment

Background: Serum albumin is widely used in clinical practice as a nutrition-related biomarker, yet in sepsis its concentration is strongly influenced by inflammation, capillary leak, fluid redistribution, hepatic dysfunction, and exogenous albumin administration. Although hypoalbuminemia is consistently associated with poor outcomes, exogenous human albumin solution (HAS) has not shown a clear survival benefit in septic patients. We therefore investigated whether the prognostic significance of early serum albumin change differs according to whether the change is spontaneous or occurs in the context of HAS administration. Methods: We analyzed septic patients from the MIMIC-IV database and assessed consistency in an independent external cohort from the First Affiliated Hospital of Xi'an Jiaotong University. Delta albumin (ΔALB) was defined as the change in serum albumin during the first 3 ICU days. Patients were stratified according to HAS use during this period. The primary outcome was 90-day survival. In the external cohort, the absence of a clear prognostic association in albumin-treated patients was directionally consistent, although validation of endogenous ΔALB was limited because most patients received HAS. Endogenous ΔALB was associated with baseline albumin, hematocrit change, platelet change, crystalloid volume, and liver-related variables, suggesting important influences of hemodilution, illness severity, circulatory status, and hepatic dysfunction. Conclusion: In the primary MIMIC-IV cohort, spontaneous increases in serum albumin, but not albumin changes among HAS-treated patients, were associated with better outcomes. These findings suggest that early albumin recovery should be interpreted cautiously in nutritional assessment, because its clinical meaning depends on albumin source and on non-nutritional physiological factors rather than nutritional status alone.