Abstract
Appropriate estimation of inpatient cost is difficult especially among the complicated cases with several complications or combination of specific treatments. We made a regression model to estimate inpatient cost per admission for hepatocellular carcinoma(HCC) with or without decompensated complications. 260 patients hospitalized due to HCC in our hospital were tracked retrospectively. For each case, the period of hospitalization and records of underwent specific medical interventions during it were extracted from the insurance claim data.
Relationship between the total costs(insurance reimbursement) included basic hospital charge and each 0-1 variable corresponding to execution of the specific medical interventions for each complication or HCC therapy(resection, percutaneous ethanol injection, Chemo-lipiodol therapy or parental anticancer agent injection, etc.) was investigated using univariate and multivariate analysis.
As a result of multivariate analysis, a model consisted of several variables of medical intervention: resection, Chemo-lipiodol therapy and parental anticancer agent injection, and complication: ascites and edema, and variceal bleeding in addition to the period of hospitalization was obtained. It predicted total inpatient cost with the high correlation coefficient(0.89) among HCC cases of separated data set for validation.
This study suggested that the model consisted of several specific medical interventions corresponding to each complication or treatment obtained from insurance claim data can reveal significant factors causing considerable variance of costs and predict more precisely cost of the complicated disease status.
