The prognostic value of using the duration of acute kidney injury in cardiac surgery: an example using two antifibrinolytics

Abstract

Previously, we reported that the addition of duration to the Acute Kidney Injury Network (AKIN) definition of acute kidney injury (AKI) is a marker for more severe kidney injury and predicts long-term mortality. We aimed to evaluate an example of the utility of adding AKI duration to the AKIN definition by comparing the historical use of aprotinin with Amicar. In a single-center observational study, we followed 4987 consecutive patients undergoing cardiac surgery between 2002 and 2007 for postsurgery AKI. Patients with a history of hemodialysis were excluded. Duration of AKI was calculated by the number of days AKI was present as defined by a > or = 0.3 (mg/dL) or a > or = 50% increase in serum creatinine from baseline or new onset of acute dialysis. Kaplan-Meier and Cox's proportional hazard modeling was conducted to evaluate 5-year mortality. Fifty-three percent of patients received Amicar (n = 2333) and 47% received high-dose aprotinin (n = 2093). Patients receiving aprotinin had evidence of more advanced disease and comorbidity and were more likely to develop AKI and have longer durations of AKI than Amicar (p < .001): 7.0 +/- 11.5 vs. 3.8 +/- 6.0 days (p < .001). Nearest-neighbor propensity matching demonstrated aprotinin had significantly worse 5-year mortality compared with Amicar (relative risk [RR] = 2.09, 95% confidence interval [CI] = 1.65-2.65). AKI duration added to the AKIN definition of AKI may provide the necessary sensitivity and specificity for evaluating renal outcomes in clinical trials.

Figure 1.

Duration of acute kidney injury (AKI) by antifibrinolytic agent. The percent of patients with AKI is graphed by the duration of kidney injury (1–2, 3–6, and ≥7 days) by type of antifibrinolytic: Amicar (gray bars) or aprotinin (black bars).

Figure 2.

Five-year mortality for propensity matched patients by antifibrinolytic agent. Kaplan Meier plotting 5-year mortality for nearest-neighbor propensity matched patients for Amicar (gray) or aprotinin (black). Log-rank test p < .001.