Group-based trajectory modeling of blood pressure reactivity and mortality in septic patients: a retrospective study of the MIMIC-IV database.

BACKGROUND

Circulatory instability is a key determinant of mortality in sepsis, and early hemodynamic response patterns largely shape the heterogeneity of the syndrome and the need for individualized resuscitation strategies. The Blood Pressure Reactivity Index (BPRI), calculated as the ratio of mean arterial pressure (MAP) to the vasoactive–inotropic score (VIS), has been proposed to quantify circulatory stability, with static measurements showing important prognostic value. However, whether dynamic BPRI trajectories can more accurately capture vascular reactivity phenotypes and further clarify the relationship between different trajectory patterns and clinical outcomes remains unclear.

METHODS

In this retrospective study using the MIMIC-IV v3.1 database, we identified adult sepsis patients who required continuous vasoactive support for at least 72 h during early ICU care. The BPRI was calculated every 12 h, and group-based trajectory modeling was applied to identify distinct 72-hour dynamic patterns. Baseline characteristics and clinical outcomes were compared across trajectory groups. The primary outcome was 28-day mortality. Secondary outcomes included ICU mortality, 7-day mortality, 90-day mortality, and ICU length of stay. Mortality outcomes were evaluated using multivariable logistic regression models, whereas ICU length of stay was analyzed using a generalized linear model. Kaplan–Meier survival curves were used to compare survival across trajectories, and stratified analyses were performed to examine the robustness and consistency of these associations.

RESULTS

Four BPRI trajectories were identified: an early high followed by a decline (Trajectory 1), a rapid and sustained increase (Trajectory 2), a persistently low and stable pattern (Trajectory 3), and a slow increase with moderate variability (Trajectory 4). Trajectory 3 exhibited the greatest illness severity and the highest 28-day (43%) and 90-day mortality (51%). Trajectory 2 showed progressive BPRI improvement and consistently favorable outcomes. After full adjustment, Trajectory 2 remained independently associated with reduced ICU mortality (OR = 0.52, 95% CI 0.27–0.95), 28-day mortality (OR = 0.56, 95% CI 0.32–0.96), and 90-day mortality (OR = 0.52, 95% CI 0.31–0.88). Trajectory 1 and Trajectory 4 were not independently associated with mortality. Kaplan–Meier curves demonstrated early and sustained separation of survival across trajectories, and subgroup analyses confirmed the robustness of Trajectory 2’s protective effect.

CONCLUSIONS

Dynamic BPRI trajectories identify distinct hemodynamic phenotypes in sepsis: only a continuously improving pattern predicts lower mortality, whereas low, declining, or fluctuating trajectories indicate high risk. Tracking BPRI dynamics may facilitate earlier recognition of adverse patterns and support individualized resuscitation. CLINICAL TRIAL REGISTRATION: Not applicable.