Association between admission hemoglobin level and prognosis in sepsis patients based on a critical care database.

This study aimed to explore the association between admission hemoglobin level and clinical outcomes in sepsis based on Medical Information Mart for Intensive Care IV(MIMIC-IV) database. It was a retrospective study. Sepsis patients in the database were included. Data extraction from MIMIC-IV was performed by PostgreSQL 9.6 software. Three different models including crude model (adjusted for none), model I (adjusted for gender and age) and model II (adjusted for all potential cofounders) were constructed. A generalized liner model and a smooth fitting curve for indicating the relationship between hemoglobin level and 30-day mortality were performed. 6249 septic patients with a 30.18% of 30-day mortality were included. With 1 g/dl increment in hemoglobin level, the values of odds ratio (OR) in crude model, model I and model II were 0.96 (95% confidential interval (CI) 0.94-0.99, P = 0.0023), 0.96 (95%CI 0.93-0.98, P = 0.0010) and 0.87 (95%CI 0.79-0.95, P = 0.0020), respectively. The smooth fitting curve indicated a non-linear relationship and the turning point was 7.2 g/dl. Compared the hemoglobin group < 7.2 g/dl, the risk of 30-day mortality significantly decreased by 32% in the hemoglobin group ≥ 7.2 g/dl (OR = 0.68, 95%CI 0.51-0.93, P = 0.0142). The non-linear relationship between admission hemoglobin level and 30-day mortality in sepsis was found. Hemoglobin supplementation might be beneficial for septic patients with hemoglobin level < 7.2 g/dl.