Knowledge Extraction from Hospital Information System: Using the Integrated Database of Electronic Medical Record and Patient Management System

Abstract

Objective: Using the integrated database in hospital information systems, this study tried to analyze inpatient hospital fees for stroke patients and evaluate the impact of patients' demographic and clinical characteristics on the fees. Methods: Inpatient charge data from patient management information system were merged to each patient's clinical data from electronic medical record. Demographic, clinical, health service utilization, and hospital fee data of 540 stroke patients were collected in the years 2000 and 2001. Inpatient charges were compared according to patient characteristics; in addition, the impact of those characteristics on hospital charges was analyzed using a multiple regression model. Results: The average length of stay (LOS) was 19.6 days and the total inpatient charge was USD 4 418. Inpatient charges were significantly higher in patients with previous stroke history, diabetes, hemorrhagic type, and severity, respectively. Simultaneously, LOS, ICU use, surgical operations, stroke type, hopeless discharge, lesion location, echocardiography and hypertension were significant factors influencing patient hospital charges (R2=0.847). LOS was the most significant factor explaining 81% of variance accounted for the charges. The strategies of reducing LOS need to be developed. Conclusion: Hospital information systems and their databases play an important role in a broad spectrum of research in clinical medicine, and can be useful to support retrospective studies and prospective clinical trials. Active utilization of the integrated database in hospital information systems is suggested for knowledge extraction. (Journal of Korean Society of Medical Informatics 13-2, 115-122, 2007)