Development of a Bayesian estimation tool to determine the optimal duration of apixaban discontinuation before a high-bleeding risk procedure.

Apixaban is a direct oral anticoagulant (DOAC). Many studies have shown that it shows high pharmacokinetic interindividual and intraindividual variability (IIV). The risk of hemorrhage is a major concern for patients treated with apixaban to undergo an operation or an invasive procedure. Due to this large pharmacokinetic variability, the current recommendations concerning the optimal duration of apixaban discontinuation before a high-bleeding risk procedure concern the general population and not a specific patient. The aim of this study was (1) to investigate by simulation the distribution of decay time of apixaban concentration and (2) to develop and validate an easy-to-use web tool to estimate the individual decay time of apixaban in a "real-life" situation. A systematic review of the literature was conducted to select the relevant pharmacokinetic models for the creation of the web tool. For each model, pharmacokinetic profiles were simulated and the time to reach concentrations below the threshold of 30 ng/ml (T30) was calculated. One of the selected models was chosen to perform a Bayesian estimation and predict the optimal duration of apixaban discontinuation before a high-bleeding risk procedure. All these results were concatenated into the PrevBleed application developed with the R Shiny package.