Dynamic BUN patterns in the ICU: risk stratification and prognosis following cardiac surgery.

BACKGROUND: Postoperative blood urea nitrogen (BUN) dynamics may reflect multisystem physiological stress after cardiac surgery, but their prognostic value remains underexplored. METHODS: This retrospective cohort study analyzed adult cardiac surgery patients from the Medical Information Mart for Intensive Care IV (MIMIC-IV) database. BUN trajectories in the first 0–96 h post-admission to the intensive care unit (ICU) after surgery were categorized by Latent class mixed models (LCMM). The study outcome was postoperative 360-day all-cause mortality. Kaplan-Meier survival analysis model and the log-rank test were used to evaluate the differences in the outcome among different trajectories. The Cox proportional hazards model was then applied to find the relationship between trajectories and the outcome, and to calculate the hazard ratio (HR). Finally, subgroup analysis was conducted to verify the stability of the results. RESULTS: A total of 1146 eligible patients were enrolled in this study, among whom 144 (12.6%) died during the follow-up period. 4 distinct trajectories were finally identified, with significant differences in postoperative 360-day all-cause mortality (log-rank p < 0.001). The Cox proportional hazards model revealed that, compared with the stable low-level trajectory (reference group), gradual increase trajectory (HR = 2.48, 95% CI: 1.70–3.60), rapid decline from high level trajectory (HR = 5.42, 95% CI 3.32–8.85), and marked critical elevation trajectory (HR = 3.25, 95% CI 1.54–6.87) were all associated with higher mortality risks. These differences persisted even after adjusting for variables in different models. In subgroup analysis the results persisted across most subgroups without any notable interaction (all p for interaction > 0.05). CONCLUSION: Early dynamic BUN patterns after cardiac surgery, better stratified patient mortality risk, and may be useful for the early risk assessment, personalized monitoring and prognostication.