Objectives: To develop a risk-prediction model for acute kidney injury (AKI) in patients undergoing vascular surgery.
Design: A retrospective cohort study.
Setting: A tertiary referral center.
Participants: Participants included 845 adult patients who underwent vascular surgery between January 3, 2003, and May 29, 2008.
Interventions: None.
Measurements and main results: The median age of patients was 72 years (interquartile range 65-80 years), and 653 patients (77%) were male. AKI developed in 258 (30.5%) patients. Patients with AKI had lower estimated glomerular filtration rates (60±21 v 72±21, p<0.001), were older (73 [68-78] years v 71 [65-80] years, p = 0.01), had a higher prevalence of hypertension (81% v 73%, p = 0.02), and were more likely to undergo emergency surgery (5% v 2%, p = 0.02). Patients with AKI also received more diuretics (p<0.001) and β-blockers (p = 0.003) prior to surgery. The multivariate AKI risk-prediction model with preoperative variables (estimated glomerular filtration rate, previous vascular interventions, use of preoperative diuretics and β-blockers, and emergency surgery) showed an area under the receiver operating characteristic curve of 0.67 (95% confidence interval, 0.628-0.710); a model with additional intraoperative variables (procedure duration, fluid balance, and plasma and platelet transfusion) had an area under the receiver operating characteristic curve of 0.72 (95% confidence interval, 0.685-0.760).
Conclusions: As AKI is a very common complication after vascular surgery, a risk-prediction model was derived to assess the likelihood of postoperative AKI. If validated in an independent cohort, this model may be used to facilitate targeted interventions in vascular surgery patients at high risk for AKI.
Keywords: acute kidney injury; fluid overload; predictive model; vascular surgery.
Copyright © 2015 Elsevier Inc. All rights reserved.