BACKGROUND: In an ongoing COPD study, some patients had long‐term overlapping prescriptions (>75th percentile of duration of overlapping prescriptions). Different approaches can be used to define episodes of use of medications in electronic record databases.
OBJECTIVES: To assess, among users of 4 COPD medications (LABA, LAMA, LABA/ICS, and LAMA/LABA), the impact on incidence rate (IR) and incidence rate ratio (IRR) of congestive heart failure (CHF) of 8 approaches to define episodes of use of these medications in the Clinical Practice Research Datalink.
METHODS: We defined episodes of use separately for the 4 COPD medications, allowing both 7‐day and 30‐day gaps between prescriptions, using 4 different assumptions: (1) 0% stockpiling (ie, disregarding any overlap between prescriptions), (2) 100% stockpiling for patients below the 75th percentile of duration of overlapping prescriptions and 50% for those above, (3) same as 2 except 75% instead of 50% stockpiling was used, and (4) 100% stockpiling. LABA was the reference for the IRR.
RESULTS: Using a 7‐day gap, the highest increase of person‐years (PY) of exposure across COPD medications was 6.0%, 7.1%, and 8.7% using 50%, 75%, and 100% vs 0% stockpiling. The highest increase in the number of events was 6.3%, 6.7%, and 8.1% using 50%, 75%, and 100% vs 0% stockpiling. The ranges for IR of CHF per 1000 PY were 19.1‐25.6 (0% stockpiling), 19.4‐25.5 (50% stockpiling), 19.2‐25.5 (75% stockpiling), and 19.0‐25.7 (100% stockpiling). The maximum difference across COPD medications comparing all scenarios to 0% stockpiling was 1.2% for the IR and was 0.85% for the IRR, both in the 50% stockpiling scenario. Using a 30‐day gap, the highest increase of PY of exposure across COPD medications was 2.3%, 2.9%, and 3.7% using 50%, 75%, and 100% vs 0% stockpiling. The highest increase in the number of events was 2.0%, 3.0%, and 5.9% using 50%, 75%, and 100% vs 0% stockpiling. The ranges for IR of CHF per 1000 PY were 18.6‐25.0 (0% stockpiling), 18.7‐25.1 (50% stockpiling), 18.7‐25.2 (75% stockpiling), and 19.0‐25.2 (100% stockpiling). The maximum difference across COPD medications comparing all scenarios to 0% stockpiling was 2.4% for the IR and was 4.0% for the IRR, both in the 100% stockpiling scenario.
CONCLUSIONS: As expected, increasing PY of exposure and number of events were observed for increasing stockpiling scenarios, with larger differences for the 7‐day gap. Although the differences in IRs and IRRs were larger in the 30‐day gap scenario, overall these were small (ie, ≤4.0%).
