Application of a drug-interaction detection method to the Korean National Health Insurance claims database

- We applied a drug interaction analysis model to an insurance claims database.
- The interaction between NSAIDs and diuretics was analyzed to examine the model.
- Diagnoses for the drug interaction were observed more frequently than expected.
- The result showed the potential of the database as a drug surveillance source.

Drug interactions (DIs) constitute a serious problem and are considered to contribute to 6-30% of all adverse events (AEs). The use of existing data, including claims data, is expected to be helpful in detecting unknown DIs by complementing conventional spontaneous reporting systems. In the present study, an 'Ω shrinkage measure' was applied to the Korean National Health claims database to test the potential of the claims database as a DI surveillance resource. A well-known DI between non-steroidal anti-inflammatory drugs (NSAIDs) and diuretics was analyzed using the model. International Statistical Classification of Diseases and Related Health Problems, 10th Revision (ICD-10) codes related to DIs were assigned to the AEs of the DIs: I50, I50.0, I50.1, I50.9, R60, R60.1, R60.9, and J81. An elevated occurrence of AEs versus the expected level was observed using a two-sided 95% lower credibility interval limit above zero, Ω025 = 0.245, which was the screening limit. The result was consistent with the actual DI between the two drugs. The finding indicates that the claims data have the potential to be used as a DI surveillance resource and that the Ω shrinkage measure may be a promising tool for detecting DIs in claims data.