Predicting acute kidney injury among burn patients in the 21st century: a classification and regression tree analysis

Abstract

Historically, acute kidney injury (AKI) carried a deadly prognosis in the burn population. The aim of this study is to provide a modern description of AKI in the burn population and to develop a prediction tool for identifying patients at risk for late AKI. A large multi-institutional database, the Glue Grant's Trauma-Related Database, was used to characterize AKI in a cohort of critically ill burn patients. The authors defined AKI according to the RIFLE criteria and categorized AKI as early, late, or progressive. They then used Classification and Regression Tree (CART) analysis to create a decision tree with data obtained from the first 48 hours of admission to predict which subset of patients would develop late AKI. The accuracy of this decision tree was tested in a separate, single-institution cohort of burn patients who met the same criteria for entry into the Glue Grant study. Of the 220 total patients analyzed from the Glue Grant cohort, 49 (22.2%) developed early AKI, 39 (17.7%) developed late AKI, and 16 (7.2%) developed progressive AKI. The group with progressive AKI was statistically older, with more comorbidities and with the worst survival when compared with those with early or late AKI. Using CART analysis, a decision tree was developed with an overall accuracy of 80% for the development of late AKI for the Glue Grant dataset. The authors then tested this decision tree on a smaller dataset from our own institution to validate this tool and found it to be 73% accurate. AKI is common in severe burns with notable differences between early, late, and progressive AKI. In addition, CART analysis provided a predictive model for early identification of patients at highest risk for developing late AKI with proven clinical accuracy.

Figure 1

Classification and Regression Tree (CART) analysis to predict late AKI. Titles above the boxes indicate predictor variables (and appropriate limits) while the boxes contain data on how each predictor split the subjects within that group according to the outcome (Late AKI). Patients who suffered late AKI are coded as black while those not suffering late AKI are gray as indicated. Terminal nodes provide the best division of the data according to the outcome of interest and could not be split further based on the Gini function and the limits of tree growth set by the investigators (see Methods section). AKI, acute kidney injury, NROF, non-renal organ failure score, Lowest 24 Hr BD, the lowest base deficit value obtained within the first 24 hours of admission, Lowest 24 Hr glucose, the lowest glucose value obtained within the first 24 hours of admission.