OBJECTIVES

: Japan has several real-word databases with different data sources. We assessed the difference between two databases by analyzing the treatment effect of oral antidiabetic drugs (OADs) in type 2 diabetes patients.

METHODS

: We used the Medical Data Vision (MDV) and Real World Data (RWD) databases. The MDV database contains health insurance claims data from DPC hospitals. The RWD database contains electronic health records and health insurance claims data from hospitals and clinics, in addition to health insurance claims data of inpatients from DPC hospitals. Both have laboratory data for patient subsets. Type 2 diabetes patients were identified with ICD-10 codes. Among the patients, those aged ≥40 years at the first diagnosis and had available HbA1c values at both the first OAD(s) prescription date (index HbA1c) and 1 year after (post-treatment HbA1c) were extracted. Simple linear regression analysis was performed between the index HbA1c and ΔHbA1c values (difference between post-treatment HbA1c and index HbA1c levels), and the regression coefficients between the two databases were compared.

RESULTS

: From the MDV and RWD databases, 11,457 and 4,531 patients were extracted. Their mean±SD index HbA1c levels were 8.1%±1.9% and 8.0%±1.9%, respectively. Among the patients, 5,013 and 2,049 had both index and post-treatment HbA1c data (mean±SD, 6.9%±1.0% vs 7.0%±1.1%, respectively). The regression equation wasΔHbA1c = 0.88 (index Hba1c − 6.6) in the MDV database, and ΔHbA1c = 0.89 (index Hba1c − 6.8) in the RWD database, which indicated that the OAD treatment reduced index HbA1c levels to <6.6% and <6.8% in 88% and 89% of the patients, respectively.

CONCLUSIONS

: We compared the MDV and RWD databases by analyzing the treatment effects of OADs. The obtained results were comparable despite the difference in data source.