Identifying medication use clusters with the R package tame based on dose, timing and ATC codes.

Area of Science:

• Pharmacoepidemiology
• Computational Statistics
• Health Informatics

Background:

• Pharmacoepidemiology often simplifies medication exposure, potentially missing crucial usage complexities.
• Existing methods may not adequately capture nuanced patterns like timing, dosage, and concurrent medication use.

Purpose of the Study:

• Introduce "tame", an R package designed for advanced medication use pattern classification.
• To provide researchers with tools to analyze complex, real-world medication data with greater accuracy.

Main Methods:

• Develop a novel distance measure within the "tame" package for clustering medication use.
• Enable customization of the distance measure using Anatomical Therapeutic Chemical (ATC) codes, timing, and dose.
• Incorporate visualization and application tools for identified medication use clusters.

Main Results:

• The "tame" package successfully identified nuanced antidepressant use patterns in a Danish pregnancy cohort.
• Demonstrated the package's ability to detect complex medication trends and improve data stratification.
• Validated the effectiveness of the bespoke distance measure in uncovering intricate medication use patterns.

Conclusions:

• "tame" offers a significant advancement in classifying medication use patterns in pharmacoepidemiology.
• The package enhances the understanding of real-world medication usage and interactions.
• Facilitates more precise patient stratification for epidemiological research.