Analysis of the Relationship between Health Conditions and Medical Expenditures Using Medical Checkup and Receipt Data

In Japan, the National Health Insurance Act was enacted in 1958, providing immense benefits to Japanese citizens by allowing all citizens to receive insured healthcare anytime and anywhere and by guaranteeing equal opportunities to receive healthcare for all citizens. However, due to various factors including changes in the demographics and growing medical needs which have become more diverse and sophisticated in recent years, medical expenditures have continued to increase and exceeded 42 trillion yen [1] in fiscal year 2016. Medical expenditures are expected to increase even further with the aging population and the advances in medical technologies. Under the current system, it is expected to become difficult to maintain the system of universal coverage, and structural reforms of medical expenditures have become a major social challenge [2].

To address this situation, efficient utilization of medical resources has become indispensable and long-term observation of the health conditions of the Japanese people as a whole, including the healthy individuals as well as those who are ill, has become a necessity. Let us suppose that at a certain hospital there are many patients with the blood-type A and few patients with the blood-type AB. From this phenomenon, we cannot conclude that persons with blood-type A are more prone to illnesses. This is because if the most common blood group were A and the least common were AB in the population (the Japanese people as a whole), it would stand to reason that there would be more patients with blood-type A than blood-type AB. However, as healthy individuals do not voluntarily go to the hospital, it would be difficult to study their health conditions, and it is a fact that in many of the countries around the world, such studies are being conducted at considerable cost.

In Japan, workers aged 40 or older are required to undergo regular medical checkups ("Designated Checkups") [3], under the Industrial Safety and Health Act. Therefore, data for healthy individuals as well as those who are ill already exist. The combined membership of the National Federation of Health Insurance Societies, the Japan Health Insurance Association and the Mutual Benefit Associations runs into the tens of millions and is therefore a massive dataset, which is larger than any dataset that has been analyzed before [4]. Furthermore, receipts from each healthcare facility are sent on a monthly basis to the health insurance societies, which hold payment-related data including information on medical treatment, medical fees, and pharmaceutical fees. In other words, a huge dataset already exists (including data of healthy individuals) that represents a larger sample than anything studied in previous studies. Under the current privately-paid treatment system, as a patient is free to choose his or her own hospital, a hospital cannot know what kind of treatment the patient had received at another hospital. Although systems exist for recording prescription drugs ("Okusuri Techo"), they do not cover all medical information. Therefore, health insurance societies are the only organizations that hold data on both the health conditions and the medical practices/treatments of its members. However, such health data are not being effectively used and much of the data is destroyed after the mandatory retention period of five years.

Two years ago, our research team launched the RIETI Research Project titled Exploring Inhibition of Medical Expenditure Expansion and Health-oriented Business Management based on Evidence-based Medicine (Leader: NAWATA Kazumitsu), and we have been conducting analytical studies on ways of efficiently using medical resources. First of all, we built a database that integrated Designated Checkup and receipt data provided by several health insurance societies. The results of the analyses have already been presented as research papers in a number of academic journals as well as two RIETI Discussion Papers [5] [6]. The following is a brief summary of these studies.

The main points of the research are as follows. The effects of lifestyle-related diseases (diabetes, hypertension, dyslipidemia and hyperuricemia) on medical expenditures were analyzed using several different models. The model known as the power transformation Tobit model was used because the data contained many zeros, i.e. individuals with no medical expenditures in the fiscal year, and otherwise the distribution was markedly skewed to the right, representing a small number of individuals with large medical expenditures during the period. As expected, if the individual had a lifestyle-related disease, the medical expenditures became higher under almost all of the models. We also observed that medications increased the expenditures significantly, beyond the price of the medications themselves, for those with lifestyle-related diseases over the period. In the case of diabetes, medical expenditures tended to surpass those for the other three lifestyle-related diseases. Furthermore, if the individual had the preexisting condition of cerebrovascular disease, cardiovascular disease or kidney failure/dialysis, medical expenditures became markedly higher. It was particularly observed that in cases where the individual had multiple lifestyle-related diseases including diabetes and kidney failure/dialysis, medical expenditures became exorbitantly high, and medical expenditures for such individuals, on average, exceeded 100,000 points, or were 13.6 times as much as those for healthy individuals. The analysis illustrated that for the efficient use of medical resources, preventing lifestyle-related diseases through proper health counseling, as well as preventing the more critical stages of the diseases through early treatment were essential.

In November 2017, the American College of Cardiology (ACC), the American Heart Association (AHA) and nine other organizations announced the new hypertension guideline (2017 ACC/AHA Guideline) [7]. According to the Guideline, the criterion for hypertension has been changed from the previous 140/90 mmHg to 130/80 mmHg. In response to this change, we conducted an empirical analysis of the relationship between blood pressure and medical expenditures in order to determine if this change in the Guideline was appropriate. We analyzed the factors that affected the distributions of blood pressure using a regression model. We found out that blood pressure was affected by several factors such as age, gender, height, BMI (body mass index) and certain lifestyles. Next, we analyzed the relationship between medical expenditures and blood pressure using the power transformation Tobit model. Although there was a positive relationship between systolic blood pressure (SBP) and medical expenditures under a simple two-variable analysis, in the power transformation Tobit model, a significant negative was found. In other words, the tendency was the higher the SBP, the lower the medical expenditures. When the explanatory variables of age, gender and BMI were added, the estimated SBP became negative, pointing toward the importance of considering the relationships between the explanatory variables. The results of this research did not support the new 2017 ACC/AHA Guideline, at least, in terms of the effects of SBP, and pointed toward the necessity of a review of research on a wide range of diseases, not limited to cardiovascular diseases such as heart disease and blood disorders, as well as studies on cost-effectiveness and new research. Europe, Canada, Japan and other countries will not adopt these guidelines [8] – [10] and will continue to follow the previous guidelines but this issue requires our full attention going forward.

We will continue to conduct analyses of the health conditions of individuals in the long term with the cooperation of an even greater number of health insurance societies. We hope that our study will contribute to the efficient use of medial resources and the creation of the ideal form of the medical insurance system.