Intradialytic Hemodynamic Instability and Delirium in Critically Ill Patients Receiving Renal Replacement Therapy: A Retrospective Cohort Study.

Abstract

Background: Acute kidney injury (AKI) requiring renal replacement therapy (RRT) is common in critically ill patients and is frequently accompanied by delirium. Intradialytic hemodynamic instability, particularly hypotension, may impair cerebral perfusion; however, its quantitative relationship with delirium remains unclear. This study investigated the association between intradialytic hemodynamic instability and Confusion Assessment Method for the Intensive Care Unit (CAM-ICU)-defined delirium in critically ill patients receiving RRT. Methods: This single-center retrospective study included 87 adult intensive care unit (ICU) patients undergoing RRT between September 2025 and May 2026. Delirium was assessed using the CAM-ICU. Intradialytic hemodynamic burden was quantified by the cumulative duration of mean arterial pressure (MAP) < 65 mmHg and the number of hypotensive episodes during RRT. Multivariable logistic regression analyses adjusted for age, sex, and admission SOFA score were performed to evaluate the associations between hemodynamic parameters and delirium, while receiver operating characteristic (ROC) analyses were used to assess their discriminative ability. Results: Of 87 patients, 46 (52.9%) were classified as CAM-ICU-positive for delirium. Patients with delirium had a longer cumulative duration of MAP < 65 mmHg (median, 36.0 vs. 5.0 min; p = 0.048) and more hypotensive episodes (median, 3.0 vs. 1.0; p = 0.032). These differences were no longer significant after normalization to the number of RRT sessions (p = 0.153 and p = 0.249). In addition, neither parameter showed a statistically significant association with delirium after adjustment for age, sex, and admission SOFA score. ROC analysis demonstrated limited discriminative ability (AUC, 0.62 and 0.63, respectively). Mechanical ventilation duration, ICU length of stay or ICU mortality did not show a statistically significant association with CAM-ICU positivity in the adjusted models. Sensitivity analyses using alternative functional forms for cumulative MAP < 65 mmHg duration, including log-transformed, categorical, and natural cubic spline models, did not show statistically significant associations with CAM-ICU positivity. Conclusions: Critically ill patients with delirium had greater cumulative intradialytic hypotensive burden; however, these differences were attenuated after accounting for the number of RRT sessions, and the hemodynamic parameters did not show statistically significant associations with delirium in the adjusted regression models. These findings suggest a potential relationship between intradialytic hemodynamic burden and delirium but do not establish an independent or causal association.