Acute kidney injury in intensive care unit patients: incidence, risk factors and mortality in a single-center retrospective cohort

Aims: Data on acute kidney injury (AKI) from state hospital intensive care units (ICUs) are limited. We determined the frequency of AKI over the ICU stay, the factors associated with it and its relationship with mortality using the risk, injury, failure, loss, end stage kidney disease (RIFLE) criteria. Methods: Adults with complete data staying at least 24 hours in a state hospital ICU between January 2022 and January 2025 were reviewed retrospectively; only index admissions were analyzed, each assigned the worst RIFLE class reached. Loss and End-stage classes, defined by dialysis dependence duration, were excluded. AKI already present at admission was counted together with AKI arising later, and exposures were recorded at any point during the stay, so the proportion reported is a period frequency of mixed prevalent and incident disease and the temporal sequence between exposure and renal injury could not be established. Because mechanical ventilation and vasopressor support were used in almost every patient who died, organ support could not serve as a covariate in the mortality model and is reported descriptively. Results: Of 890 patients (median age 76 years), 521 (58.5%) had AKI: risk 205 (23.0%), injury 141 (15.8%), failure 175 (19.7%). ICU mortality rose stepwise: 18.4%, 34.6%, 63.8% and 80.0% (p for trend <0.001). Diuretic infusion [odds ratio (OR) 2.88, 95% confidence interval (CI) 1.75-4.74], nephrotoxic drug exposure (OR 2.37, 1.50-3.72), mechanical ventilation (OR 1.92, 1.19-3.10) and illness severity (OR 1.06 per APACHE II point, 1.03-1.09) were independently associated with AKI. After adjustment for age, illness severity, sepsis, malignancy and admission category, the adjusted odds of death were 1.30 (0.82-2.05) for risk, 3.13 (1.87-5.24) for injury and 6.32 (3.76-10.64) for failure. Conclusion: AKI affected nearly three-fifths of this elderly, high-acuity cohort and mortality rose with RIFLE severity. An association with death independent of age, illness severity, sepsis, malignancy and admission category was demonstrable at the Injury and Failure thresholds but not at Risk; because organ support could not be adjusted for, these are upper bounds on the contribution of renal dysfunction rather than point estimates. The factors identified are associations rather than demonstrated causes, since the timing of exposure relative to renal injury could not be established, and organ support was so nearly universal among those who died that its contribution could not be separated from the outcome itself.