OBJECTIVES: Generally, cost parameters used in partitioned survival model (PSM) are summarized as follows; (1) drug costs in pre-progression, (2) other costs in pre-progression, (3) drug costs in post-progression, and (4) other costs in post-progression. The purpose of this study was to develop a method to estimate these cost parameters by using claims database from the perspective of the Japanese healthcare system.

METHODS: Japan Medical Data Center (JMDC) Claims database was used to estimate these parameters. Receipt data of lung cancer patients was extracted by using ICD-10 code. The number of patients and total person-time was 10,883 and 119,409 person-month (Jan 2005 - Jan 2017), respectively. Cost data with binary variables which indicate the use of lung cancer drugs were converted to panel dataset. In this study, we considered eight lung cancer drugs. The month in which the lung cancer drug first appeared was taken as the treatment start month and the month in which it appeared last was assumed to be the treatment end month due to progression of disease. Panel data analysis was performed to estimate cost parameters in pre-progression and post-progression in each lung cancer drug.

RESULTS: Average monthly drug costs for pre-progression ranged from JPY 46,149 to JPY 2,463,270. Drug costs for post-progression were estimated from JPY 130,876 to JPY 1,135,732. Other costs for pre- and post- progression ranged from JPY 108,450 to JPY 453,855, and from JPY 145,009 to JPY 456,490, respectively.

CONCLUSIONS: Although there are several technical problems, claims database is considered to be useful data source in cost-effectiveness analysis based on PSM. Further research is required on a method to identify occurrence of clinical event based on claim database.