Lactate trajectories and outcomes in patients with sepsis in the intensive care unit: group-based trajectory modeling

Wei Yu, Jinqiang Zhuang, Jiaqi Li, Zixuan Wang, Jing Wang, Xiaojie Zhang, Junling LengFrontiers in Public Health 2025 · DOI

Abstract

In this study, different subtypes of sepsis were successfully identified by analyzing lactate trajectories. Combined with the dynamic changes in lactate levels, the Group-Based Trajectory Modeling (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.

Introduction

Sepsis is a life-threatening organ dysfunction caused by a dysregulated host response to infection and remains a leading cause of mortality in intensive care units (ICUs) worldwide. A significant challenge in managing sepsis is its extreme heterogeneity; patients sharing the same diagnosis often exhibit vastly different biological states and clinical trajectories. While static "snapshot" measurements of blood lactate are commonly used to gauge disease severity, they fail to capture the dynamic physiological response to treatment. This study explores whether longitudinal patterns, or trajectories, of lactate levels can more accurately identify distinct patient subphenotypes and predict 28-day mortality.

Methods

The study utilized the Medical Information Mart for Intensive Care (MIMIC-IV) database, analyzing a cohort of 1,575 adult patients who met Sepsis-3 criteria and had daily lactate measurements for at least five days. Researchers employed Group-Based Trajectory Modeling (GBTM), a specialized application of finite mixture models designed to identify clusters of individuals following similar longitudinal trends. Missing data (under 30%) were handled using Multiple Imputation by Chained Equations (MICE) with the Classification and Regression Trees (CART) algorithm. Model selection was guided by the Bayesian Information Criterion (BIC) and Akaike Information Criterion (AIC). Survival outcomes were assessed using Kaplan–Meier curves and Cox proportional hazard models.

Results

Initial analysis revealed a clear divergence in lactate trends between survivors and non-survivors. While both groups often start with elevated levels, survivors typically show a more pronounced decline over the first five days of ICU admission.

Trajectory of lactic acid change over time between the survival group and the death group.

The GBTM analysis further refined this by identifying five distinct subphenotypes based on lactate clearance patterns:

  1. Low-stable: Consistently low levels (Lowest mortality: 23.2%).
  2. Low-slowly declining: Low baseline with a gentle downward trend.
  3. High-rapidly decline: High baseline but rapid clearance.
  4. Moderate-slow declining: Mid-range baseline with slow clearance.
  5. High-slow decline: High baseline with persistent elevation (Highest mortality: 46.8%).
Lactate trajectories based on GBTM.

A key finding was that the rate of lactate clearance (the "slope") was often more predictive than the initial value. For instance, the "high-rapidly decline" group had significantly better outcomes than the "high-slow decline" group despite similar starting points. Furthermore, lactate levels from days 2 through 5 were more strongly associated with mortality risk than the initial day 1 measurement, suggesting that persistent hyperlactatemia after initial resuscitation is a critical warning sign.

Discussion

This study demonstrates that lactate trajectories provide a powerful tool for risk stratification in sepsis. By identifying subphenotypes with distinct mortality risks, clinicians can move toward more personalized interventions. However, the study has limitations: it is a retrospective, single-center analysis based on the MIMIC-IV database and requires multi-center prospective validation. Additionally, the five-day window focuses on the acute phase and may not capture late-stage complications. Future research should explore higher-frequency data and real-time monitoring to further enhance the clinical utility of trajectory modeling.

Summary updated

Abstract

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.
BibTeX
@inproceedings{yu2025lactate,
  author = {Wei Yu and Jinqiang Zhuang and Jiaqi Li and Zixuan Wang and Jing Wang and Xiaojie Zhang and Junling Leng},
  title = {{Lactate trajectories and outcomes in patients with sepsis in the intensive care unit: group-based trajectory modeling}},
  booktitle = {Frontiers in Public Health},
  year = {2025},
  doi = {10.3389/fpubh.2025.1610220},
}
Figure