Abstract
BACKGROUND: Sepsis is highly heterogeneous. Therefore, identifying biomarkers that can stratify patients with sepsis into more homogeneous cohorts to develop individualized treatment and care measures for patients and carry out early intervention to reduce the risk of death and improve the prognosis of patients has become a current research hotspot. METHODS: Using the MIMIC-IV database, we analyzed data from 1,575 adult patients with sepsis. Serum lactate levels were measured once daily for 5 consecutive days after admission. The GBTM model was used to stratify the risk of sepsis and explore the relationships between different lactate trajectories and 28-day mortality in septic patients. RESULTS: We report a new method for identifying subphenotypes of sepsis patients based on lactate trajectories. Through group-based trajectory modeling, we identified and validated five groups of sepsis patients with different lactate trajectories, namely, "Low-stable group," "low-slowly declining group," "high-rapidly decline group," "Moderate-slow declining group," and "high-slow decline group." The relationships between sepsis patients with different lactate trajectories and 28-day mortality were explored. Among them, patients with a "Low-stable group" had the lowest in-hospital 28-day mortality. Patients with a "high-slow decline group" had the highest 28-day mortality. CONCLUSION: In this study, different subtypes of sepsis were successfully identified by analyzing lactate trajectories. Combined with the dynamic changes in lactate levels, the GBTM model was used to stratify patients according to their risk of sepsis. This model provides a theoretical basis for clinicians to evaluate the prognosis of patients using the lactate change trajectory.