Prediction of acute kidney injury in the immediate postoperative period following liver resection: a retrospective cohort study.

Purpose: Following liver resection, acute kidney injury (AKI) is a frequent and potentially reversible complication. No predictive model of postoperative AKI following liver resection has assessed the role of oliguria in the postanesthesia care unit (PACU). Our objectives were 1) to estimate the association between PACU oliguria and AKI and to develop and internally validate a predictive model of postoperative AKI using variables available in the PACU and 2) to assess the additive predictive value of PACU urine output.

Methods: We conducted a retrospective cohort study of patients who underwent elective liver resection. Our primary outcome was the occurence of AKI within seven days after surgery. We used two sets of candidate predictors (17 and 11 variables, respectively) to develop a predictive model for postoperative AKI, including PACU urine output. We first calculated risk ratios (RR) with 95% confidence intervals (CIs) for different definitions of PACU oliguria. We then fitted multivariable logistic regression predictive models with a least absolute shrinkage and selection operator and reported optimism-corrected model performance properties.

Results: We included 1,520 patients. Both the incidence of AKI within seven days after surgery and the prevalence of PACU oliguria, defined as a urine output < 0.5 mL·kg^-1·hr^-1, were 11%. PACU oliguria was associated with AKI (RR = 1.74; 95% CI, 1.20 to 2.50). Both predictive models had good discrimination (area under the receiver operating characteristic curves, 0.775 and 0.766, respectively) and excellent calibration. PACU urine output increased models' discrimination.

Conclusion: Oliguria in the PACU was associated with AKI within seven days after surgery. We developed predictive models for 7-day AKI following liver resection that had good performance properties.