A methodological framework for constructing opioid agonist therapy episodes in administrative health data.

Area of Science:

• Health Informatics
• Pharmacovigilance
• Data Science

Background:

• Opioid Agonist Therapy (OAT) is crucial for reducing overdose risk.
• Administrative health data are increasingly used to study OAT outcomes.
• Existing methods for OAT episode construction have limitations in capturing treatment complexity.

Purpose of the Study:

• To develop a robust framework for constructing OAT episodes in administrative health data.
• To more accurately reflect real-world OAT use, including medication switches and concurrent therapies.
• To improve the precision of OAT outcome studies using health records.

Main Methods:

• Utilized Allen's interval algebra, temporal margins, and drug-specific permissible gaps to define OAT episodes.
• Analyzed OAT dispensations from people living with HIV in British Columbia (2010-2020).
• Characterized temporal relations between dispensations (e.g., meets, before, overlaps) to differentiate monotherapy, transitions, and multitherapy.

Main Results:

• The 'before' relation was most common for OAT, especially methadone (99.43%).
• The 'equals' relation was notable for buprenorphine (21.46%) and slow-release oral morphine (20.87%).
• Identified 10,386 episodes: 93.53% monotherapy, 3.72% transition, 2.27% multitherapy, and <1% transition-multitherapy.

Conclusions:

• Proposed a novel episode-building framework using interval algebra, temporal margins, and permissible gaps for OAT data.
• This framework enhances the classification of OAT use in administrative health data.
• The method is adaptable for use in other settings and populations with parameter adjustments.