Associations of Heart Rate Trajectories with Mortality and AKI Occurrence in Septic Patients: A Retrospective Study from MIMIC-IV and eICU Databases

Background: Abnormal heart rate is a common early manifestation of sepsis, but the association of the dynamic changes with clinical outcomes remains unclear. This study aimed to investigate the relationship between dynamic heart rate trajectories and 28-day mortality as well as acute kidney injury (AKI) in patients with sepsis. Methods: This retrospective cohort study was based on two independent cohorts from the eICU-CRD and MIMIC-IV databases, enrolling 7537 patients with sepsis. Group-based trajectory modeling (GBTM) was used to identify dynamic heart rate trajectories within the first 7 days after ICU admission. Landmark analysis with a landmark time of day 7 to account for immortal time bias. Cox regression and logistic regression analyses were applied to evaluate the associations of different trajectories with 28-day mortality and new-onset AKI risk after the 7-day landmark. Sensitivity analyses confirmed the consistency with the primary analysis. Results: Five distinct heart rate trajectories were identified: persistently low-stable (Traj-1), moderate-level gradually increasing (Traj-2), high-baseline rapidly declining (Traj-3), low-level U-shaped (Traj-4), and persistently high (Traj-5). Using Traj-1 as the reference, among patients observed at day 7, patients with Traj-5 had elevated risks of 28-day mortality (all p < 0.05) and new-onset AKI after the 7-day landmark (all p < 0.05). Conclusions: Persistently high heart rate trajectories (Traj-5) were associated with higher risks of 28-day mortality and new-onset AKI among patients observed at day 7. These findings support careful heart rate monitoring in septic patients can provide valuable information into the prognostic association and risk stratification, while residual confounding precludes causal interpretation.