Mapping Partners Master Drug Dictionary to RxNorm using an NLP-based approach

Area of Science:

• Health Informatics
• Natural Language Processing
• Pharmacology

Background:

• Medication terminology mapping is crucial for healthcare interoperability.
• Existing methods for mapping local drug dictionaries to standardized terminologies like RxNorm can be labor-intensive.
• Automating this process supports meaningful use initiatives in healthcare.

Purpose of the Study:

• To develop and evaluate an automated natural language processing (NLP) method for mapping a local medication terminology (Partners Master Drug Dictionary) to RxNorm.
• To facilitate the creation and maintenance of accurate RxNorm mappings for improved data exchange and clinical decision support.

Main Methods:

• A natural language processing (NLP) tool, MTERMS, was employed to map 5961 terms from the Partners Master Drug Dictionary (MDD) and 99 top prescribed medications to RxNorm.
• Mapping was performed at both term and concept levels.
• A gold standard, created by domain experts through manual review, was used to assess the NLP tool's performance.

Main Results:

• The NLP tool achieved high precision (99.8% for all MDD terms, 100% for top 99 terms) in mapping to RxNorm.
• Recall for the NLP tool was 73.9% for all MDD terms and 72.6% for the top 99 terms.
• An exact semantic match to RxNorm was found for 74.7% of MDD terms and 82.8% of the top 99 terms.

Conclusions:

• Mapping challenges between local drug dictionaries and RxNorm stem from unique representation requirements and differing modeling approaches.
• An automated NLP approach, augmented by human expert review, offers an efficient and feasible solution for dynamic medication terminology mapping.
• This methodology supports enhanced interoperability and the meaningful use of electronic health data.