Association between platelet-albumin-bilirubin score and 30-day in-hospital mortality in patients with sepsis: evidence from a large database and machine learning modeling

The three components of the platelet-albumin-bilirubin (PALBI) score are closely related to inflammatory responses, hepatic dysfunction, and coagulation abnormalities in sepsis, but its prognostic significance in this population remains unclear. This study aimed to examine the association between PALBI score and 30-day in-hospital mortality in patients with sepsis. Data from the Medical Information Mart for Intensive Care (MIMIC-IV) database were retrospectively analyzed. Patients diagnosed with sepsis were included in this study. The primary outcome was 30-day in-hospital mortality. Multivariate Cox regression analyses were performed to assess the association between PALBI score and mortality after adjusting for potential confounders. Subgroup analyses were also performed to assess the robustness of the study results. Exploratory machine-learning analyses were additionally performed to assess the relative contribution of PALBI among candidate variables. A total of 6,636 participants were included. PALBI scores were categorized as ≤ -2.53, -2.53~-2.09, and >-2.09. A linear association between PALBI scores and mortality was observed after adjustment for confounders. In the fully adjusted model, each 1-point increase in PALBI was associated with a higher risk of 30-day in-hospital mortality (HR: 1.373, 95% CI: 1.269,1.485, p < 0.001). In exploratory SHAP-based analyses, PALBI was identified as a contributory feature. High PALBI scores were independently associated with higher 30-day in-hospital mortality in ICU patients with sepsis. PALBI may serve as an objective and readily available adjunctive marker for early risk stratification. However, its standalone predictive performance is limited.