In Japan, rapid population aging has called for the provision of contiguous medical and long-term care services to the elderly. In this situation, the Ministry of Health, Labor and Welfare is promoting the establishment of an Integrated Community Care System that can appropriately provide various daily life support services. This system requires cooperation of relevant organizations such as hospitals, nursing facilities, and pharmacies in an area. In order to measure the state of cooperation and identify the issues in establishing the Integrated Community Care System, a questionnaire survey is executed to these organizations. However, there is no clarity on an appropriate design of the questionnaire survey, which can accurately identify the issues. As a result, there were problems with the conventional questionnaire survey, and thus, the results obtained by the conventional questionnaire survey cannot be utilized to identify such issues.
The purpose of this study is to propose a design method of the questionnaire survey, which can identify issues accurately. First, we analyzed the conventional questionnaire and survey results to clarify problems. Thereafter, we considered the measures to overcome these problems and proposed an appropriate design for questionnaire survey. We confirmed that the questionnaire was improved by the proposed method.
The Taguchi's (2005) T-method belongs to the MT (Mahalanobis Taguchi) system, which is a representative method in quality engineering. The T-method is used for prediction or estimation in fields such as economics and weather forecasting.
Masuda (2012) pointed out that the T-method has the advantage of analyzing missing data, which was not available in previous multiple regression analysis methods. The T-method integrates the results of single regression analysis. Therefore, it can analyze data including missing data without processing the missing data. When the sample size is small and multiple items are missing, the previous single imputation method may not complement the missing value in some cases.
Therefore, this study aims to propose the T-method along with improvements to analyze data including missing values without deleting them. Then, we propose a new single imputation method using the calculation procedure, and compare it with the previous single imputation method.
Infectious diseases pose a significant threat to humankind. Specifically, influenza is one of the most threatening infectious diseases in Japan, infecting large numbers of people and greatly influencing the community. To mitigate this impact, it is important for health organizations to predict and identify outbreaks and to take appropriate measures to keep the infection from spreading. Usually, infected patients would visit the pharmacy after getting a medical consultation. Therefore, it is hypothesized that pharmacy sales data will reflect the general health condition of local residents. However, most studies on predicting disease outbreaks are based on sentinel surveillance data; information on the relationship between pharmacy sales data and disease outbreaks is less well known. In this study, the sales data from several pharmacies in the Tochigi prefecture were evaluated, and the results were compared with surveillance data. The SIR model, which is a classical model in which disease dynamics is represented by differential equations, was then used to predict the number of people infected with influenza in the Tochigi prefecture over a period of three seasons. Focusing on the peak day, the difference was 8 days in the most accurately predicted season, thus demonstrating the usefulness of using pharmacy sales data.