Methodology — How Our Data Works

Data Sources

Our primary data source is the OpenFDA API, which provides structured access to FDA drug labeling data — including the Drug Interactions sections of FDA-approved labels, which document known interactions between medications.

Drug identification is standardized using RxNorm, the National Library of Medicine's standardized nomenclature for clinical drugs. This ensures that different brand names and generic names for the same medication are correctly matched.

Extraction Process

FDA drug labels contain drug interaction information in unstructured prose text. We use natural language processing (NLP) and pattern matching to extract individual drug-drug interaction pairs from this text, along with their severity, mechanism, and clinical effect when described.

Each extracted interaction is validated against the source label and mapped to standardized drug identifiers (RxCUI) using the RxNorm API. Interactions that cannot be mapped to specific drugs are excluded from the pair-level database. Our current database covers 257,000+ FDA drug labels and 22,000+ unique drug-pair interactions.

Severity Classification

Interaction severity is classified into four levels based on the language used in FDA drug labeling:

Contraindicated

The FDA label explicitly states the drugs should not be used together. Typical language: "contraindicated," "do not use concomitantly," "must not be combined."

Major

The interaction may cause serious clinical consequences. Typical language: "serious," "significant," "potentially dangerous," "life-threatening."

Moderate

The interaction may require monitoring or dosage adjustment. Typical language: "monitor," "use with caution," "may increase/decrease levels."

Minor

The interaction is unlikely to cause significant clinical effects. Typical language: "minor," "not clinically significant," "small effect."

Update Frequency

Our data pipeline runs periodically to incorporate new and updated FDA drug labels. The OpenFDA dataset is updated regularly as the FDA approves new drugs and revises existing labels. Each interaction record on this site references the source FDA label.

Known Limitations

No drug interaction database is complete. FDA labels do not always describe interactions in a structured format, so some interactions may not be captured by our extraction process. Severity classifications are derived from label language and may not always align with clinical severity scales used by other databases such as Lexicomp or Micromedex. Our database focuses on drug-drug interactions and does not comprehensively cover drug-food, drug-alcohol, or drug-condition interactions.

Corrections and Feedback

If you identify an error in our data or a missing interaction, please contact us at info@checkdruginteractions.com. We investigate all reported issues and update our database accordingly.