Computer aided warfarin dosing in the Swedish national quality registry AuriculA - Algorithmic suggestions are performing better than manually changed doses.

INTRODUCTION: Warfarin treatment with a high time in therapeutic range (TTR) is correlated to fewer complications. The TTR in Sweden is generally high but varies partly depending on local expertise and traditions. A dosing algorithm could minimize variations and increase treatment quality. Here we evaluate the performance of a computerized dosing algorithm.
MATERIALS AND METHODS: 53.779 warfarin treated patients from 125 centers using the Swedish national quality registry AuriculA. If certain criteria are met, the algorithm gives one of seven possible dose suggestions, which can be unchanged, decreased or increased weekly dose by 5, 10 or 15%. The outcome evaluated by the resulting INR value was compared between dose suggestions arising from the algorithm that were accepted and those that were manually changed. There were no randomization, and outcomes were retrospectively analyzed.
RESULTS: Both the algorithm-based and the manually changed doses had worse outcome if only two instead of three previous INR values were available. The algorithm suggestions were superior to manual dosing regarding percent samples within the target range 2-3 (hit-rate) or deviation from INR 2.5 (mean error). Of the seven possible outcomes from the algorithm, six were significantly superior and one equal to the manually changed doses when three previous INR:s were present.
CONCLUSIONS: The algorithm-based dosing suggestions show better outcome in most cases. This can make dosing of warfarin easier and more efficient. There are however cases where manual dosing fares better. Here the algorithm will be improved to further enhance its dosing performance in the future.