Early arterial lactate trajectories and mortality risk in critically ill heart failure: a two-cohort trajectory analysis

Lactate is widely used as a biomarker of tissue hypoperfusion and illness severity in critically ill patients with heart failure (HF). While static lactate levels have prognostic value, dynamic changes in lactate over time may offer deeper insights into metabolic stress and clinical outcomes. However, the prognostic utility of lactate trajectories remains poorly defined in HF populations. We conducted a retrospective cohort study using the MIMIC-IV (n = 5,261) and MIMIC-III (n = 906) databases to identify distinct early arterial lactate trajectories in ICU-admitted HF patients. Latent class mixed model were used to categorize 72-hour lactate patterns, and association with in-hospital, 28-day, and 1-year mortality were assessed using multivariable logistic and Cox regression models. External validation was performed in the MIMIC-III cohort. Three distinct lactate trajectory classes were identified: low-stable (Class 1, 86.4%), early rise with delayed decline (Class 2, 4.1%), and early decline followed by re-elevation (Class 3, 9.6%). Compared with Class 1, Class 2 had higher in-hospital mortality (OR 6.88, 95% CI 4.86-9.74), 28-day mortality (HR 3.88, 95% CI 3.17-4.75), and 1-year mortality (HR 3.16, 95% CI 2.65-3.78; all P < 0.001). Class 3 also showed higher risks versus Class 1 (OR 3.03, 95% CI 2.32-3.98; 28-day HR 2.20, 95% CI 1.83-2.66; 1-year HR 1.83, 95% CI 1.56-2.15; all P < 0.001). Risks showed a consistent gradient (Class 2 > Class 3 > Class 1) across cohorts. Findings were consistent in the validation cohort. Sensitivity and subgroup analyses confirmed the robustness of these associations. Early arterial lactate trajectories were independently associated with both mortality in critically ill patients with HF. Trajectory-based profiling provides more nuanced prognostic insight than initial lactate values alone and may inform early risk stratification and ICU decision-making.