Association of Albumin Infusion With Differential Response in Pediatric Sepsis and Septic Shock: Retrospective Analysis Using a U.S. Multicenter 2012-2018 Dataset.

OBJECTIVE The study goal was to evaluate the outcomes associated with albumin use in children with sepsis and shock compared with those without shock, using causal inference analysis in a multicenter cohort.

DESIGN, SETTING, AND PATIENTS This was a secondary analysis of electronic health record data collected from 13 U.S. PICUs between 2012 and 2018, consisting of children younger than 18 years who met Phoenix sepsis criteria within the first 24 hours of PICU admission.

INTERVENTIONS Covariate-balancing propensity score weighting was applied to adjust for indication bias in the albumin use. Patients receiving at least 0.5 g/kg albumin within 24 hours of PICU admission were assigned to the albumin group; others to the control. Only 24-hour survivors were included to address immortal time bias.

MEASUREMENTS AND MAIN RESULTS Overall, 17,307 children with sepsis survived at least 24 hours. Of these, 1,344 patients (7.8%) who received albumin within the first 24 hours, and 9,678 (55.9%) met the criteria for septic shock. A significant interaction between albumin use and shock status was observed (interaction: -0.353, p = 0.007), with albumin administration in pediatric septic shock patients associated with lower in-hospital mortality: odds ratio equals to 0.698 (95% CI, 0.629-0.774), risk ratio equals to 0.746 (95% CI, 0.625-0.891), and hazard ratio equals to 0.688 (95% CI, 0.558-0.848). In contrast, there was no difference in outcomes between the albumin and control groups in the non-shock group.

CONCLUSIONS Early albumin administration was associated with improved outcomes in children with septic shock, but not in those without shock. These results highlight the importance of considering clinical heterogeneity, such as the presence of shock, in identifying treatment-responsive subgroups and enabling more targeted interventions in pediatric sepsis. Further prospective validation is warranted.