BACKGROUND: In the context of approval of the atrial fibrillation (AF) indication for dabigatran etexilate, as a follow-up measure the sponsor and EMA agreed to evaluate potential off-label use of dabigatran in Europe. The methodological challenges of using data sources with different types of clinical data had not been studied in detail.
OBJECTIVES: Discuss estimated prevalence of potential off-label use and associated methodological challenges.
METHODS: Observational, cross-sectional study on dabigatran in three databases with different types of clinical information available: Cegedim Strategic Data Longitudinal Patient Database (CSD-LPD), France (cardiologist panel, n=1,706 [FR-1]; general practitioner panel, n=2,813 [FR-2]; primary care information); National Health Databases, Denmark (n=28,619; hospital episodes, dispensed ambulatory medications [DK]); and Clinical Practice Research Datalink (CPRD), UK (linkable to Hospital Episode Statistics [HES], n=2,150 [UK-1]; not linkable, n=1,285 [UK-2]; hospital and primary care data were available for HES-linkable patients) (Aug 2011-Aug 2015). Two definitions were applied to estimate potential off-label use based on either recorded diagnoses or proxies: a broad definition of on-label prescribing using codes for disease indication (e.g., AF) and a restrictive definition excluding patients with conditions for which the drug is not indicated (e.g., valvular AF).
RESULTS: Key methodological challenges: availability of detailed hospital and primary care clinical information impacted results observed across data sources. Limited information available likely led to overestimation of off-label use, particularly in CSD-LPD, and may explain the disparate prevalence estimates across countries. Estimates under the broad definition: UK-1 5.7%, UK-2 11.5%, DK 17.1%, FR-1 24.1%, FR-2 34.0%; and under the restrictive definition: UK-1 17.4%, UK-2 25.6%, DK 29.1%, FR-1 37.5%, FR-2 44.1%. No diagnoses potentially related to anticoagulant use could be identified in a large proportion of potential off-label users.
CONCLUSIONS: Results need to be interpreted cautiously due to limitations in the availability of data (no primary care data in Denmark; no hospital data in France). In this context, CPRD HES-linkable estimates are likely to be the most accurate. Availability of detailed clinical data is crucial for this type of research.
