(203) Medicine utilisation studies in Australian individual-level dispensing data: A blinded, multi-centre replication experiment

Background: Medicine dispensing data require extensive preparation when used for research and different decisions made by different analysts during this process may lead to results that do not replicate.

Objectives: To conduct an experiment examining the impact of analytical decisions on results of a study measuring discontinuation, intensification and switching in a cohort of patients initiating metformin.

Methods: This was a prospective meta-research study. Four Australian sites independently developed a HARmonized Protocol Template to Enhance Reproducibility (HARPER) protocol based on the same master protocol and executed their analyses using the same Australian Pharmaceutical Benefits Scheme 10% sample dataset of medicine dispensing records from 2015 to 2019. These data include the strength and quantity of medicine dispensed, but not the intended dose or duration of treatment. Each site calculated cohort size and demographics, estimated periods of exposure, and operationalised treatment events including discontinuation, switch to another diabetes medicine, and intensification (addition of another diabetes treatment). Sites then estimated one-year incidence and time to event for each treatment event type, and used Cox proportional hazards regression models to estimate associations between cohort characteristics and time to each treatment event type. Concordance was assessed by measuring deviations from the calculated median of each value across the sites.

Results: Good agreement was found across all sites for the number of initiators (median: 53,127, range: 51,848-55,273), gender (median: 56.9% female, range: 56.8-57.1%) and age group. Each site employed different methods for estimating periods of exposure and used different operational definitions for the treatment events. Consequently, poor agreement was found for the one-year incidence of discontinuation (median: 55%, range: 34-67%), switching (median: 3.5%, range: 1-7%), intensification (median: 8%, range: 5-12%), time to event estimates and hazard ratios.

Conclusions: Differences in analytical decisions when deriving exposure from dispensing data affect replicability. Harmonised approaches to defining exposure periods and standardised concepts of treatment events will enhance the quality of medicines utilisation research.