To classify sepsis patients based on blood urea nitrogen (BUN) trajectories and to examine their association with 28-day mortality.

Based on BUN measurements taken on the 1st, 3rd, and 7th days after admission, we employed a latent growth mixture modeling (LGMM) to identify trajectory classes. Additionally, we plotted Kaplan-Meier survival curves and performed log-rank tests, followed by the application of multivariable Cox proportional hazards models to evaluate associations with 28-day mortality. Prespecified subgroup analyses were also conducted.

A total of 210 patients diagnosed with sepsis were included in this study. Based on the longitudinal trajectories of BUN, three distinct trajectory patterns were identified: a low-initial slow-decline trajectory, a high-initial persistently high-level trajectory, and a moderate-initial slow-decline trajectory. The corresponding 28-day all-cause mortality rates for these trajectories were 20.0% (25/125), 64.0% (16/25), and 46.7% (28/60), respectively. Kaplan-Meier survival analysis demonstrated clear separation of the 28-day survival curves among the three groups, with a statistically significant difference observed between groups (log-rank χ² = 20.030, P < 0.001). In comparison to the low-initial slow-decline trajectory, patients in both the high-initial persistently high-level trajectory and the moderate-initial slow-decline trajectory demonstrated significantly lower overall survival probabilities; notably, the high-initial persistently high-level group exhibited the lowest 28-day survival rate, indicating the poorest 28-day prognosis (all P < 0.001). Cox regression analysis showed that, using the low-initial slow-decline trajectory as the reference, patients classified in the high-initial persistently high-level trajectory [hazard ratio (HR) = 2.500, 95% confidence interval (CI) (1.200, 5.400), P = 0.020] and those in the moderate-initial slow-decline trajectory [HR = 2.600, 95%CI (1.400, 4.700), P = 0.002] exhibited a significantly increased risk of 28-day all-cause mortality.

Dynamic BUN trajectories delineate prognostically heterogeneous subtypes of sepsis, with the persistently high trajectory exhibiting the highest mortality rate. Continuous monitoring of BUN levels can facilitate early risk assessment and enable risk-stratified management strategies.