(C-081) An Algorithm for Mapping the Treatment Sequences of Patients with COPD: Development and Application in a Canadian Cohort.

Background: Acute exacerbations of chronic obstructive pulmonary disease (COPD) are linked to poor outcomes but may be prevented through inhaled maintenance therapy with long-acting β2 agonists (LABA), long-acting muscarinic antagonists (LAMA), and inhaled corticosteroids (ICS). The dynamic nature of COPD therapy makes delineating treatment sequences in database studies challenging.

Objectives: To describe an algorithm for identifying and mapping COPD therapy sequences, as applied in a study describing the treatment trajectories of Canadian patients.

Methods: We identified patients aged ≥40 years with a prescription for a first (index) maintenance inhaler from 2018 to 2024 and a prior COPD diagnosis in provincial claims from British Columbia. We developed an algorithm to identify patients' inhaled therapy regimens during each of their first four treatment lines, out of six options: LABA, LAMA, LABA/ICS, LABA/LAMA, LAMA/ICS, and LABA/LAMA/ICS, each including ≥1 separate inhalers. All identified patient prescriptions were used to create continuous treatment ‘episodes’ for each inhaler type, ending when a ‘conflicting’ episode (sharing an ingredient) began. The algorithm determines first-line regimen based on inhalers filled within 14 days of the index inhaler. It then determines the regimen for the next line using two key dates: A) the earliest episode end date for inhalers in the current regimen and B) the earliest episode start date for non-regimen inhalers. If A < B, the algorithm identifies the latest end date of all inhalers in the current regimen. If this date is less than B, or B does not exist, it assigns the patient as having discontinued current therapy. Otherwise, the patient is assigned a second-line regimen based on the available inhaler episodes at that time. If instead, B < A, the patient is assigned a second-line regimen based on all inhalers on hand 14 days after B.

Results: The cohort included 29 865 patients, all of whose treatment sequences were successfully mapped. The most common first-line regimen was LAMA (39.7%) followed by LABA/ICS (38.5%), LABA/LAMA (12.6%), LABA/LAMA/ICS (7.2%), LABA (1.2%) and LAMA/ICS (0.8%). Patients in the LABA/LAMA group had the longest mean treatment duration (267.2 days, SD = 363.9), while LAMA/ICS had the shortest (120.1 days, SD = 148.2). There were 25 836, 16 606, and 14 228 patients in the second, third, and fourth lines, respectively. Treatment discontinuation rates were 75.5%, 41.2%, and 76.0% in the first, second, and third lines, respectively. Across all lines, ICS-inclusive regimens accounted for the most discontinuations.

Conclusions: Our algorithm delineated complex COPD treatment trajectories, highlighting the dynamic nature of COPD therapy. This approach may inform similar efforts to map treatment sequences.