Multiple-center evaluation of mortality associated with acute kidney injury in critically ill patients: a competing risks analysis

INTRODUCTION In this study, we aimed to assess the association between acute kidney injury (AKI) and mortality in critically ill patients using an original competing risks approach. METHODS Unselected patients admitted between 1997 and 2009 to 13 French medical or surgical intensive care units were included in this observational cohort study. AKI was defined according to the RIFLE criteria. The following data were recorded: baseline characteristics, daily serum creatinine level, daily Sequential Organ Failure Assessment (SOFA) score, vital status at hospital discharge and length of hospital stay. Patients were classified according to the maximum RIFLE class reached during their ICU stay. The association of AKI with hospital mortality with "discharge alive" considered as a competing event was assessed according to the Fine and Gray model. RESULTS Of the 8,639 study patients, 32.9% had AKI, of whom 19.1% received renal replacement therapy. Patients with AKI had higher crude mortality rates and longer lengths of hospital stay than patients without AKI. In the Fine and Gray model, independent risk factors for hospital mortality were the RIFLE classes Risk (sub-hazard ratio (SHR) 1.58 and 95% confidence interval (95% CI) 1.32 to 1.88; P < 0.0001), Injury (SHR 3.99 and 95% CI 3.43 to 4.65; P < 0.0001) and Failure (SHR 4.12 and 95% CI 3.55 to 4.79; P < 0.0001); nonrenal SOFA score (SHR 1.19 per point and 95% CI 1.18 to 1.21; P < 0.0001); McCabe class 3 (SHR 2.71 and 95% CI 2.34 to 3.15; P < 0.0001); and respiratory failure (SHR 3.08 and 95% CI 1.36 to 7.01; P < 0.01). CONCLUSIONS By using a competing risks approach, we confirm in this study that AKI affecting critically ill patients is associated with increased in-hospital mortality.