The Characteristics of Individual-Based Linked Data Based on Japanese Medical Infrastructure Law

Speaker(s)

Shoji A1, Kudo K2, Murashita K2, Nakaji S2, Igarashi A3
1The University of Tokyo, Bunkyo-ku, 13, Japan, 2Hirosaki University, Hirosaki, Aomori, Japan, 3The University of Tokyo, Bunkyo-ku, Tokyo, Japan

OBJECTIVES: To explore health issues, data from before the onset of diseases is needed. However, in Japan the sharing of data generated in hospitals with third parties is strictly limited. The next-generation medical infrastructure law (NGMIL) enacted in 2018 is one available framework. The aim of this study is to confirm the advantage and disadvantage of data linked under the NGMIL framework.

METHODS: We compared the characteristics of residents in Hirosaki City using individually-linked medical and check-up data under the NGMIL with data from a collaboration Hirosaki City, Hirosaki University, and the University of Tokyo. The former identified each resident using birthday and Insurance identification number (IIN) allocated by the Hirosaki City Office before anonymization. The latter used only IIN which can be changed for several reasons. In addition, the former deleted data of residents who have rare diseases and who opted out. We identified residents who had 6-month continuous enrollment before 1 July (baseline period) in 2015 and figured the age distribution, coronary heart disease (CHD) risk scores, and Charlson Comorbidity Index (CCI) during the baseline period.

RESULTS: We identified 46383 and 66436 subjects who had 6-month continuous enrollment in 2015 in the NGMIL and collaboration data set, respectively. In the former, after identifying unique residents, the number of subjects decreased. Subjects were younger in the collaboration data (mean±SD, 58.7±19.3 and 52.2±20.9 for the NGMIL and collaboration data, respectively), because the change in IINs causes censoring. The CHD risk score and CCI were similar between data (39.3±8.6 and 1.0±1.7, 39.2±9.0 and 0.5±1.3, respectively), suggesting that deleting data of severe disease patients and residents who opted out would have only a limited impact on future analyses using the NGMIL data.

CONCLUSIONS: We confirmed that the NGMIL data has a great potential benefit and few limitations.

Code

RWD40

Topic

Epidemiology & Public Health, Real World Data & Information Systems, Study Approaches

Topic Subcategory

Data Protection, Integrity, & Quality Assurance, Health & Insurance Records Systems, Public Health

Disease

Cardiovascular Disorders (including MI, Stroke, Circulatory)