Using a Semiautomated Procedure (CleanADHdata.R Script) to Clean Electronic Adherence Monitoring Data: Tutorial

Area of Science:

• Digital Health
• Health Informatics
• Medication Adherence Research

Background:

• Patient adherence to medications is crucial and can be monitored using digital health technologies like electronic monitors (EMs).
• Understanding changes in treatment and deviations in EM use is essential for accurate adherence assessment.
• Characterizing these factors is key to establishing true medication adherence levels.

Purpose of the Study:

• To introduce and provide a user guide for the CleanADHdata.R computer script.
• The script is designed to clean raw electronic monitor (EM) adherence data.
• This tool aims to standardize and improve the analysis of digital health adherence data.

Main Methods:

• Collected raw EM data alongside adherence start/stop dates, prescribed regimens, and patient demographics.
• Identified expected daily EM openings and deviations from prescribed use.
• The script formats data longitudinally and calculates daily medication implementation.

Main Results:

• A simulated dataset for 10 patients using 15 EMs over a median of 187 days was analyzed.
• Median patient implementation improved from 83.3% to 97.3% after data cleaning (a 14% increase).
• This substantial improvement highlights the script's capability to prevent data misinterpretation.

Conclusions:

• The CleanADHdata.R script offers a semiautomated approach, enhancing standardization and reproducibility in adherence data analysis.
• This tool has wide-ranging applications for cleaning adherence data from various digital health technologies.
• It provides a cleaned dataset crucial for accurate adherence analysis, including implementation and persistence.