Background: Universal agreement on criteria for acute renal failure (ARF) is lacking. The purpose of the current study was to determine which of 6 definitions for ARF best predicted clinical outcomes in postoperative cardiothoracic surgery (CTS) patients.
Methods: Criteria for ARF were retrospectively applied to 1,085 CTS patients. General linear models analyzed length of stay (LOS) and ventilator days with logistic regression for mortality.
Results: Thirty-seven percent of patients met at least 1 of 6 definitions of ARF. For each 1-mg/dL increase from the initial creatinine, LOS increased by 6.96 days, ventilator days increased by 3.58 days, and mortality increased by 2.23 times (P < .0001).
Conclusions: One definition that best predicted ARF was not found. ARF was a significant independent predictor of increased mortality, LOS, and ventilator days. Even small increases in creatinine correlate with clinically significant worsening of expected outcomes.