새로운 환자표본자료 표본 추출 및 대표성 검증

Abstract

Background: The sampling framework of the National Patient Sample of Health Insurance Review & Assessment Service is needed to be improved due to the current demographic structure. We proposed a sampling method and additional strata for extracting the National Patient Sample data due to the current demographic structure, such as low birth rate and aged population. Methods: A total of 36 strata were set by adding four strata compared to the existing one. The maximum rate of minimal sample number was defined among the entire strata. Based on the rate, we extracted a small-scale sample dataset consisting of about 400,000 people and a large-scale sample dataset of more than 700,000 people. Results: The representativeness of the high-frequency disease and the low-frequency disease was confirmed. For health expenditure, the representativeness of samples was confirmed in large-scale samples. However, the representativeness of small-scale samples was not confirmed in five strata. Conclusion: Using the maximum rate of minimal sample number can reflect the demographic structure changes and diverse medical utilization. Although lack of representativeness in the five strata of the small-scale sample, both the small-scale sample and the large-scale sample are necessary to improve data accessibility and a sustainable data provision system. It will be helpful in establishing health policies and conducting medical research.