Back to Annual Meeting Posters

Background: Abnormal preoperative laboratory values have been associated with increased mortality in patients undergoing hepatic resection for malignancy. However, cutoff values for these preoperative labs have been defined arbitrarily. The aim of the current study was to identify cut off values for these preoperative laboratory denominators which can help identify patients at increased risk of mortality.
Methods: Patient undergoing liver resection for malignancy (primary and secondary) were extracted from 2005-2010 National Surgical Quality Improvement Database. We determined the optimal cutoffs for each laboratory denominator using the classification and regression tree analysis (CART), and the "party" package for conditional inference trees in R. Patients were classified according to the cutoffs determined from CART analysis and logistic regression analysis was used to fit a multivariate model, with backward variable selection.
Results: A total of 4812 patients who underwent liver resections for malignancy were included. Statistically significant association was seen between increased 30 day mortality and preoperative laboratory values including serum Na <= 135 meq/L, BUN >19 mg/dl, serum creatinine >1.68 mg/dl, serum albumin <=2.6 g/dl, bilirubin >1.8 mg/dl, SGOT >50 IU/L, alkaline phosphatase of > 149 IU/L, WBC >10,790/ul , Hct <= 28, and INR >1.1. In a multivariate logistic regression model, albumin <=2.6, SGOT>50, INR>1.1, BUN>19, and alkaline phosphatase>149 are independent predictors of 30 day mortality with an area under the curve of 0.77.
Conclusions: Cutoff laboratory values defined in the current study may help identify patients who are at higher risk of mortality from hepatic resections.