A rational approach for selection of optimal covariate-based dosing strategies

At present, there is no rational approach for choosing a dosing strategy for individualization based on a covariate. An approach to use in establishment of an a priori dosing strategy for individualization is presented. Factors influencing the choice of such a dosing strategy are identified.

The approach requires definition of the following: target variable, seriousness of deviations from the target (ie, risk function), population model, covariate distributions, and constraints. Minimizing the total risk yields an optimal dosing strategy, estimated as dose sizes for different subpopulations and covariate cutoff values at which doses are increased or decreased. The method was illustrated with the use of simulated and real drug examples for the situation in which clearance is related to creatinine clearance.

The estimated optimal cutoff(s) paralleled the median creatinine clearance in the population. The extent of variability in clearance explained by creatinine clearance was the main factor influencing the optimal ratios between adjacent dose sizes. An optimal dosing strategy was possible to estimate for the real drug.

The method is simple to perform, although one difficulty lies in defining the target variable and risk function. Our results imply that commonly used constraints in dosing strategies based on renal function (ie, dose ratio of 2 and predetermined cutoffs) are nonoptimal in the sense we propose. Because an optimal dosing strategy may not be practical to use, the therapeutic cost that would result with any constraint can be assessed by comparison of the outcome after the desired and the optimal strategy.