BACKGROUBD: Roux-Y gastric bypass (GB) reduces obesity related comorbidity, but considerable health care resources are consumed performing this operation. Greater understanding of the factors influencing operative costs is needed to permit rational cost-benefit analyses by health care providers and insurors. Our aim was to determine to what extent clinical variables can be used to predict GB hospital costs and to develop a model for these costs.
METHODS: From 7/97 through 8/00, 161 patients underwent GB by a single surgeon at a teaching hospital. Clinical variables were age, sex, BMI, and presence of diabetes, hypertension, sleep apnea, heart failure, laparoscopic approach, and the complications leak, PE, MI, SBO, GI bleed, decubitus wounds, stenosis, wound infection, postop transfusion, and incisional hernia. Total hospital costs (TC) were the summation of fixed and variable costs of individual charge items derived from inpatient billing records. Stepwise multiple regression was used to derive a linear model of TC both with and without TC outliers (5) included.
RESULTS: 123 women and 38 men underwent GB with an average (mean±SD) age of 43±8 yrs and BMI of 56±12 kg/m2, and 40% were laparoscopic. Comorbidities included: DM 25%, HTN 54%, sleep apnea 57%, heart failure 9%. Mortality within 3 months occurred in 1.2%. Complications occurred in 35%, but most occurred in the outpatient setting not reflected in TC. In-hospital complications occurred in 9% including leak 1.2%, MI 1.2%, PE 3%, SBO 1.2%, GI bleed 2.5%, and decubitus wounds 1.9%. The TC (mean±SEM) was ±1808 including outliers and ±258 excluding outliers. Regression including outliers yielded a four variable model including DECUB, MI, leak, and transfusion with an adjusted R2 of .60. With outliers excluded, a six variable model yielded an adjusted R2 of .43: TC = 7797 + 68*age + 8214*MI + 2958*transfusion + 9669*decubitus + 3529*GI bleed + 3949*SBO BMI showed colinearity with decubitus wounds and was a significant predictor of TC only when postop complications were not included in the model.
CONCLUSIONS: Significant TC variability (43%) can be attributed to patient age and the occurrence of a small number of high cost complications including decubitus wounds, MI, transfusion, and leak. This analysis can be used to estimate the cost of complications after GB. Larger amounts of health care resources will likely be consumed when performing GB on older patients and on those with larger BMI.
