BACKGROUND: : There is a wide spectrum of disease burden in hepatocellular carcinoma accompanied by several options for surgical management. However, the associated mortality of such procedures is not well defined. Accurate predictions of patients' perioperative risk would be helpful to guide decision making.
METHODS: : The Nationwide Inpatient Sample was queried for data from 1998 to 2005. A cohort of patients who were discharged for hepatic procedures with a diagnosis of primary liver neoplasm was assembled. Procedures were categorized as hepatic lobectomy, wedge resection, or enucleation/ablation. Logistic regression and bootstrap methods were used to create an integer risk score for estimating the risk of in-hospital mortality using procedure type, patient demographics, comorbidities, and hospital type. A randomly selected sample of 80% of the cohort (n = 2263) was used to create the score with validation conducted in the remaining 20% (n = 571).
RESULTS: : In total, 2834 patient discharges were identified. Overall in-hospital mortality was 6.52%. Factors that were included in the final model were age, sex, Charlson comorbidity score, procedure type, and teaching hospital status. Integer values were assigned to these characteristics and were used to calculate an additive score. Four clinically relevant score groups were assembled to stratify the risk of in-hospital mortality, with a 19-fold gradient of mortality that ranged from 1.5% to 28.3%. In the derivation set, as in the validation set, the score discriminated well with c-statistics of 0.75 and 0.73, respectively.
CONCLUSIONS: : The current results indicated that an integer-based risk score can be used to predict in-hospital mortality after surgery for hepatocellular carcinoma, and it may be useful for preoperative risk stratification and patient counseling. Cancer 2010. (c) 2010 American Cancer Society.
Available at: http://works.bepress.com/jennifer_tseng/63/