MO026
IMPROVED ACUTE KIDNEY INJURY (AKI) STAGING WITH SERUM CREATININE TRAJECTORIES

INTRODUCTION AND AIMS: AKI is an important complication in the inpatient setting and is associated with poor short-term (inpatient death, need for renal replacement therapy (RRT) and long-term outcomes (end-stage renal disease (ESRD) and/or death). The risk for inpatient mortality associated with AKI is clearly associated with its severity, as judged by the increases in serum creatinine (sCr). The current consensus staging criteria (KDIGO) is based on the extent of increase in sCr from baseline, while analysis of sCr trajectories provides the opportunity to develop AKI staging based on the rate of rise of sCr. Our purpose was to compare the discrimination between stages and association with mortality for slope-based trajectories and KDIGO AKI staging.

METHODS: We analyzed initial admissions for 53,122 patients at UAB Hospital between October 2009 and September 2013, excluding patients with length of stays <1 or >30 days, with prior dialysis or kidney transplantation, with <3 sCr determinations, or with minimum sCr below 0.4 mg/dL. Individualized AKI Thresholds were calculated for each patient using a modified Bayseian approach, and reflected the 95% confidence intervals around the time-course of all sCr values (sCr trajectory). AKI Episodes were defined by sCr values that exceeded the AKI Threshold, and the slopes were calculated for each episode using the peak sCr value and the most proximate minimal sCr value and these criteria: >3 sCr available values used for slope regression, change in sCR >0.3 mg/dL, and elapsed time between minimum and peak sCr values >8 hours. These slopes were clustered into 3 non-overlapping groups, using the STATA kmeans command, defining sloped-based Stages 1-3. The reference group (Stage 0) were subjects without any AKI Episodes.

RESULTS: The distribution of Deaths and Patients by Stages are show for KDIGO Staging and Slope-Based Staging in the table:

Discrimination was significantly improved with Slope-Based compared to KDIGO_Stages (Net Reclassification Improvement = 0.207 (95% CI: 0.168_0.246, P<0.001). Risk of inpatient mortality significantly increased with the number of AKI Episodes (1 episode, 874 (13%) deaths for 6,534 patients; 2 episodes, 170 deaths( 23%) for 728 patients; 3 episodes, 24 deaths (35%) for 69 patients. Distinct AKI episodes during an admission were not recognized by KDIGO-based AKI staging.

CONCLUSIONS: Sloped-Based Staging of inpatient AKI outperforms the current consensus KDIGO staging for assessing the risk of inpatient mortality, and also permits risk assessment for the number of AKI episodes during and admission. Serum creatinine slope-based trajectories provide a real-time risk assessment tool for outcomes, while KDIGO staging is based on the maximum sCr excursion and is only applicable to risk-assessment during the course of an admission.