Objective: In order to understand characteristics of atrial fibrillation patients in the era of new oral anticoagulants (NOACs), this study explores differences in characteristics between patients treated with dabigatran etexilate (DE) and warfarin (W) that may be due to patient channeling in 'real-world' clinical practice.
Research design and methods: Medco claims data were used to characterize 41,805 non-valvular atrial fibrillation (NVAF) patients from the US with a DE (N = 7055) or W (N = 34,750) prescription between February 2011 and April 2012. The first prescription for each treatment in this period defined the index date. The treatment groups were stratified by newly diagnosed or warfarin-experienced patients. Characteristics, comedications, and comorbidities in the 12 month period prior to index date were assessed.
Results: Newly diagnosed patients initiating DE had overall lower use of comedications compared to W patients. In contrast, warfarin-experienced patients switching from W to DE showed higher use of antibiotics, beta blockers, gastrointestinal drugs and NSAIDs compared to patients remaining on W. Newly diagnosed NVAF patients initiating DE showed lower proportions for comorbidities such as myocardial infarction, congestive heart failure, and renal disease. This was also reflected in the Charlson comorbidity index (CCI) (mean DE 2.1 vs. W 3.0) and the CHA2DS2-VASc score (mean DE 3.4 vs. W 4.0). For warfarin-experienced NVAF patients, these differences were not seen. Interpretation of results is limited by the fact that administrative claims data are not gathered for scientific research. Underreporting of non-serious conditions might occur and life-style variables, laboratory values and over-the-counter medication were not available.
Conclusions: As also seen for other newly marketed drugs, differences in baseline characteristics, comedication, and comorbidities were detected between DE and W in newly diagnosed patients, as well as in warfarin-experienced patients. This channeling may have significant impact on comparative outcome studies if not properly addressed in study design and analysis.