Daniel F.B. Wright (1), Stephen B. Duffull (1)
(1) School of Pharmacy, University of Otago, Dunedin, New Zealand
Objectives: Warfarin is a difficult drug to dose accurately and safely due to large inter-individual variability in dose requirements. Current dosing strategies achieve INRs within the therapeutic range only 40-65% of the time [1,2]. Bayesian forecasting methods have the potential to improve INR control. The aims of this study were (1) to assess the predictive performance of a Bayesian dosing method for warfarin implemented in TCIWorks, and, (2) to determine the expected ‘time in the therapeutic range’ (TTR) of INRs predicted using TCIWorks.
Methods: Patients who were initiating warfarin therapy were prospectively recruited from Dunedin Hospital, Dunedin, New Zealand. Warfarin doses were entered into TCIWorks from the first day of therapy until a stable steady-state INR was achieved. The KPD model developed by Hamburg et al was used as the prior model [3]. The Bayesian method was used to predict steady-state INRs (INRss); (1) under the prior model, .i.e. without using any observed INR values to individualise parameters, then, (2) by incorporating the first measured INR value, i.e. the posterior prediction of INRss based on 1 INR value, then, (3) by incorporating the first and second measured INR values, i.e. the posterior prediction of INRss based on the first 2 INR values (4) and then by sequentially including additional measured INR values. Observed and predicted INRss values were compared using measures of bias (mean prediction error [MPE]) and imprecision (root mean square error [RMSE]) [4]. The TTR was determined by calculating the percentage of predicted INRss values between 2 and 3
Results: A total of 55 patients completed the study. The prior model was positively biased (MPE 0.52 [95% CI 0.30, 0.73]) with an RMSE of 0.96. The bias became non-significant (MPE -0.09 [95% CI -0.23, 0.05]) and imprecision improved (RMSE <0.54) once 3 or more INR values were entered into TCIWorks. Based on the level of imprecision in the prediction of INRss, the Bayesian dosing method was able to predict the INRss within the therapeutic range 65% of the time when 3 INR values were included and 70-80% of the time with 4-6 INR values.
Conclusions: The Bayesian warfarin dosing method produced accurate and precise steady-state INR predictions after the inclusion of 3 or more INR values. The expected TTR of 65-80% is a substantial improvement in INR control compared to current dosing methods. Further evaluation of this method in the clinic is warranted.
References:
[1] van Walraven C, Jennings A, Oake N, et al. Effect of study setting on anticoagulation control. Chest 2006; 129 (5): 1155-66.
[2] Witt DM, Sadler MA, Shanahan RL, et al. Effect of a centralized clinical pharmacy anticoagulation service on the outcomes of anticoagulation therapy. Chest 2005; 127: 1515-1522.
[3] Hamberg AK, Wadelius M, Lindh JD, et al. A pharmacometric model describing the relationship between warfarin dose and INR response with respect to variations in CYP2C9, VKORC1, and age. Clin Pharmacol Ther 2010; 87 (6): 727-34.
[4] Sheiner LB, Beal SL. Some suggestions for measuring predictive performance. J Pharmacokinet Biopharm 1981 Aug; 9 (4): 503-12.
Reference: PAGE 22 () Abstr 2699 [www.page-meeting.org/?abstract=2699]
Poster: Other Drug/Disease Modelling