Validation of patient‐reported warfarin dose in a prospective incident cohort study

Abstract

Inconsistencies in the definition and the collection of warfarin dosing data could lead to bias in observational, clinical, and pharmacogenetic studies. The present study aims to assess the concordance between patient-reported and prescribed warfarin doses among new warfarin users in the Quebec Warfarin Cohort (QWC) study. Demographic, clinical, and lifestyle data were collected at cohort entry and each three months during a 1-year follow-up period among a subgroup of 219 patients from the prospective QWC study. We evaluated the differences between reported and prescribed warfarin doses overall and at each follow-up period. Concordance was tested in a multivariate generalized linear mixed model and allowed to vary from 95% to 105% of the prescribed dose. Overall, there was no significant difference between reported and prescribed warfarin doses (p>0.05, Pearson coefficient=0.969, power=100%). There was also no significant difference across each of four timepoints tested (p>0.05). We found that 84.0% of the reported warfarin doses were concordant with the prescribed doses. Having a history of myocardial infarction was significantly associated with a low concordance (OR=0.494; CI 95%: 0.286-0.852). In our population, we found that patient-reported warfarin dose and prescribed warfarin dose were comparable for the conduct of observational and clinical studies as well as for the validation and implementation of warfarin dosing algorithms. Moreover, the effect was similar whether measured in new-onset users of warfarin and after up to 12 months of use.