High-Throughput Computing to Detect Harmful Drug-Drug Interactions in Older Adults: Protocol for a Population-Based...

Area of Science:

• Pharmacoepidemiology
• Health Informatics
• Geriatric Medicine

Background:

• Drug-drug interactions (DDIs) pose significant risks, particularly for elderly individuals on multiple medications.
• Traditional methods for detecting harmful DDIs are slow and often miss critical interactions.
• Existing population-based studies face challenges in identifying DDIs due to the vast number of potential drug combinations.

Purpose of the Study:

• To outline a novel protocol for efficiently identifying harmful drug-drug interactions (DDIs) in older adults.
• To leverage administrative health care data for robust DDI detection.
• To inform safer prescribing practices and regulatory decision-making.

Main Methods:

• Conducting population-based, new-user cohort studies using Ontario's linked administrative health care data (2002-2023).
• Utilizing high-throughput computing to analyze cohorts of Ontario residents aged 66+ years.
• Employing propensity score methods and regression models to evaluate 74 acute outcomes within 30 days, controlling for over 400 baseline characteristics.

Main Results:

• Preliminary analysis identified 3.8 million older adults and over 500 unique medications, enabling study of approximately 200,000 potential drug combinations.
• Initial drug pair cohorts showed a median of 583 new users, with a median overlap of 57 days.
• The study protocol was finalized in August 2025, with analysis completion scheduled for fall 2026.

Conclusions:

• This study will identify credible signals of harmful DDIs in older adults within routine care settings.
• The innovative approach integrates high-throughput computing and rigorous pharmacoepidemiologic methods for real-world evidence generation.
• Findings are expected to enhance medication safety and support informed clinical and regulatory decisions.