Ingredient-based method to create medication lists and support granular data segmentation.

Area of Science:

• Pharmacology
• Health Informatics
• Computational Biology

Background:

• Accurate medication lists are crucial for patient care and research.
• Existing methods for generating medication code sets can be labor-intensive and prone to errors.
• The need for automated, reliable methods to create up-to-date medication lists is increasing.

Purpose of the Study:

• To demonstrate the generalizability of a previously developed ingredient-based method for automated medication list creation.
• To apply and validate this method in the National Drug Code (NDC) context for generating opioid and antidepressant medication lists.
• To assess the accuracy, completeness, and up-to-dateness of the automatically generated lists.

Main Methods:

• Reused and applied an ingredient-based methodology in the National Drug Code (NDC) context.
• Validated generated medication lists via automatic comparison with curated and active medication databases (CDC, HEDIS, FDA, RxNorm).
• Included manual physician review as part of the validation process.

Main Results:

• The ingredient-based method proved generalizable across two clinical terminologies (RxNorm and NDC).
• High accuracy, reusability, and completeness were demonstrated for both opioid and antidepressant medication code sets.
• The method successfully generated up-to-date and error-free medication lists.

Conclusions:

• Automated, ingredient-based methodologies are effective for creating accurate and comprehensive medication lists.
• This approach offers an innovative alternative to less accurate, data-driven methods for generating clinical lists.
• Such methodologies are vital for supporting patient privacy regarding sensitive medication information.