Assessment of Medication Safety Using Only Dispensing Data.

Area of Science:

• Pharmacoepidemiology
• Health Informatics
• Drug Safety Research

Background:

• Medication dispensing databases are valuable for large-scale drug utilization studies, especially in universal healthcare systems.
• These data can identify inappropriate medicine use linked to safety concerns but often lack diagnosis codes and confounder data.
• Adverse drug event (ADE) detection is challenging due to the absence of coded outcomes in dispensing data.

Purpose of the Study:

• To review studies generating evidence on medicine safety using only medication dispensing data.
• To explore methods for overcoming limitations in dispensing data for medication safety assessment.
• To evaluate the utility of dispensing data in identifying potential drug safety signals.

Main Methods:

• Review of published studies on medication safety using dispensing data.
• Exploitation of medication initiation as a proxy for adverse drug events (ADEs).
• Application of self-controlled study designs, such as prescription sequence symmetry analysis (SSA), to mitigate confounding.

Main Results:

• Medication initiation can serve as a proxy for ADEs, enabling safety signal generation from dispensing data.
• Self-controlled designs, particularly SSA, effectively address unmeasured confounding in within-person comparisons.
• Dispensing data, when analyzed with appropriate methods, can contribute to medication safety assessments.

Conclusions:

• Medicine initiation as an ADE proxy is a viable strategy for safety assessments with dispensing data.
• Careful validation of proxy-event relationships and consideration of time-varying confounding are crucial.
• Data-mining dispensing data holds promise for enhancing safety assessments, though unmeasured confounding requires further investigation.