An Analysis of Medical Expenditures using Medical Checkups and Receipts using the Power Transformation Tobit Model (Japanese)

In Japan, medical expenditures have been increasing rapidly and exceeded 42 trillion yen in fiscal year 2015. It is expected that medical expenditures will continue to increase due to aging population and expenses related to advancements of medical technologies. In order to handle this situation, the efficient use of medical resources is absolutely necessary. To achieve this goal, we need to observe the health conditions of all Japanese residents including not only sick individuals but also healthy ones on a long-term basis. In general, due to the fact that healthy individuals do not go to hospitals or clinics voluntarily, it is usually difficult to learn about their health status. To remedy this, special surveys that involve massive research funding efforts have been conducted in various countries. The numbers of surveyed individuals were at most tens of thousands and survived items were limited. In Japan, most workers 40 or older are required to undergo medical checkups once a year regardless of the patient's concern for their health, in accordance with the Industrial Safety and Health Act, and therefore a huge dataset of tens of millions of Japanese individuals exists; which is a much larger sample than studied in previous surveys. In this paper, we used the power transformation Tobit Model to analyze the database for medical expenditures by combining health checks and receipts for a total of 15580 cases. Especially, we used four different models (Model A-D) to analyze the effects of four major lifestyle-related diseases（diabetes, hypertension, dyslipidemia and hyperuricemia）on medical expenditures. For general covariates other than the four lifestyle-related diseases, we found that coefficients of age, anamnesis, objective symptoms, cardiovascular diseases, kidney failure/dialysis, weight change within the previous year and night-time snacking were positive and significant at the 1% level in all models. Coefficients of gender, systolic blood pressure (SBP), LDL cholesterol, smoking, walking and not eating breakfast more than 3 times per week became negative and significant at the 1% level. Coefficients of BMI, GOT value, and willingness to consult with health professionals became positive and significant at the 5% level. For SBP and LDL cholesterol, we obtained results that were opposite to previous findings. Concerning lifestyle-related diseases, medical expenditures became higher in most models, as expected. We also observed that medical expenditures became higher if an individual was taking medications for these diseases. In the case of diabetes, medical expenditures became higher than the cases of the other three lifestyle-related diseases. Furthermore, medical expenditures became much higher when these individuals had cerebrovascular, cardiovascular diseases and kidney failure/dialysis. Especially in cases where individuals had more than one life-style related disease, including diabetes and kidney failure/dialysis, medical expenses soared; average medical expenses exceeded 100,000 points and 13.6 times as much as that of a healthy individual. For the efficient use of medical resources, it is very important to prevent life-style related diseases though proper health counseling.. Early treatments are also important to prevent an individual from suffering from more serious stages of diseases.