J Korean Soc Med Inform.  2000 Jun;6(2):1-16.

Hospital Database Marketing Using Discharge Summary DB: An Application of Data Mining Technique

• Kim YM
• Kang SH
• Kim WJ

Abstract

The environment of hospital management is becoming more turbulent as the government continuously puts pressure on hospitals to reduce medical expenditures while the public increasingly demands improved services. To survive in this turbulent environment, hospitals need to develop and apply various management techniques. Database marketing is considered as one of the useful tools. This paper applies the concept of database marketing by way of using discharge summary DB, and attempts to segment the customers (patients) to develop a differentiated marketing strategy for the target market. In particular, discharge summary DB for 1996-1999 of a university hostital located in Pusan, Korea, was utilized. The data were restructured for our study purpose and analyzed through the use of data mining technique. Major results are as follows: 1) Applying 'Decision Tree Model' suggests that the most important variable in determining the readmission(re-use of the hospital) was 'disease group. 2) Analyzing disease structure of readmitted patients reveals that the readmission rate was highest for the neoplasma patients; therefore, for this hospital, neoplasma patients can be considered as the target market. 3) Conducting cluster analysis on this target market shows that the neoplasma patients are classified into 4 groups: (1) 'Group I' can be named as 'sustaining customer(patient) group' in that they have visited the hospital until the most recent period. (2) 'Group II' is 'loyal customer(patient) group' in that frequency of admission and length of hospitalization are highest among the four groups. (3) 'Group III' can be named as 'departing customer(patient) group' in that the period between the discharge and the last visit of the hospital is shortest, which implies that they probably have changed the hospital or used other types of health care services after discharge. (4) 'Group lV' is 'average customer(patient) group' in that they show average loyalty in terms of frequency of admission, length of hospitalization, and the period between discharge and last visit.

Keyword

Database marketing; Discharge summary DB; Patient segmentation; Target patients; Data mining

MeSH Terms

Busan
Data Mining*
Delivery of Health Care
Health Expenditures
Hospitalization
Humans
Korea
Marketing*