OBJECTIVES : Warfarin’s complex dosing is a significant barrier to measurement of its exposure in observational studies using population databases, creating a significant knowledge gap. We aimed to develop a valid method to estimate patients’ individualized warfarin dose and the number of days that warfarin prescription supply will last (days’ supply), a crucial variable in studies of drug exposure.

METHODS : Using population-based administrative data from British Columbia, Canada, we developed a method based on statistical modeling, (“REWarDS”), that involved fitting a random effects linear regression model to patients’ cumulative dose over time. Model parameters included a minimal universally-available set of variables from prescription records to estimate patients’ individualized average daily dose of warfarin and days’ supply. REWarDS estimates were validated against a reference standard (days’ supply calculated using the administration instructions entered by the dispensing pharmacist) and compared to alternative methods (fixed window, fixed tablet, Defined Daily Dose, and reverse Wait Time Distribution) using repeated measures Pearson’s correlation coefficient (r), intraclass correlation coefficient (ICC) and Root Mean Squared error (RMSE), at the prescription level.

RESULTS : REWarDS-estimated days’ supply showed excellent correlation and agreement with the reference standard [r=0.90 (0.90, 0.90, p<0.0001); ICC= 0.95 (0.94, 0.95, p<0.0001); RMSE= 8.24 days] and performed better than all the alternative methods. REWarDS also provided accurate estimates for prescriptions with very short or long days’ supply, and within the first four months of therapy initiation when dose variations are the most frequent.

CONCLUSIONS : REWarDS estimated days’ supply was valid and more accurate than all other available methods. REWarDS is expected to confer optimal validity in studies using administrative data where estimates of warfarin daily dose or exposure duration are needed, such as “real-world” studies on warfarin adherence, persistence studies, clinical outcomes, and management.