Fluid Balance in the 24 Hours Before Continuous Renal Replacement Therapy and ICU Mortality Among Critically Ill Adult Patients: A Nursing Risk Assessment Study Using MIMIC-IV

Vollständiger Abstract

Worum geht es in dieser Arbeit?

Objective: To investigate the predictive value of cumulative fluid balance measured within 24 hours before initiation of continuous renal replacement therapy (CRRT) for Intensive Care Unit (ICU) mortality among critically ill adult patients. Methods: A retrospective cohort study used the MIMIC-IV 3.1 database to include adult patients who received CRRT during their first ICU admission and whose CRRT initiation occurred between ICU admission and discharge. The index time point was defined as the start of the first CRRT session. Fluid balance over the preceding 24 hours was calculated as the difference between fluid input and output and was standardized by body weight (mL/kg). Fluid balance values were winsorized at the 1st and 99th percentiles and entered into a logistic regression model in 10 mL/kg increments. The primary outcome was ICU mortality. Multivariable models assessed the association between fluid balance and ICU mortality, and model discrimination was evaluated using the C-statistic. Sensitivity analyses were performed for in-hospital mortality, specific subgroups, and after inclusion of lactate and creatinine levels. Results: Of 1,355 initially screened patients, 946 met inclusion criteria, and 444 died in the ICU (46.9% mortality rate). The body weight-standardized fluid balance during the 24 hours before CRRT was higher in non-survivors than in survivors [33.42 (16.91, 65.22) mL/kg vs. 22.71 (10.18, 45.05) mL/kg, P < 0.001]. ICU mortality rates increased across quartiles of fluid balance, from 34.2%, 44.1%, 53.4%, to 56.1% (trend test P < 0.001). Multivariable logistic regression showed that each 10 mL/kg increase in fluid balance was associated with a 5% higher risk of ICU mortality (OR = 1.05, 95% CI 1.02–1.09, P = 0.001); the C-statistic for the main model was 0.681. After adding lactate and creatinine, the C-statistic improved to 0.735, but the direction of the fluid balance effect reversed (OR = 0.94, 95% CI 0.91–0.98, P = 0.006). Conclusion: A higher body weight-standardized fluid balance within 24 hours before CRRT is associated with ICU mortality in critically ill adults and may serve as a risk signal before CRRT initiation. However, the reversal of this association when adjusting for disease severity and clinical decision-making factors suggests that its predictive value may be influenced by illness severity, treatment choices, and confounding variables. Therefore, it cannot yet support causal inference or guide fluid management independently.