Estimation of dosing strategies aiming at maximizing utility or responder probability, using oxybutynin as an example drug.

Methods for optimizing dosing strategies for individualization with a limited number of discrete doses, in terms of maximizing the expected utility of treatment or responder probability, are presented. The optimality criteria require models for both beneficial and adverse effects that are part of the utility definition and published population models describing those effects for oxybutynin (urge urinary incontinence episodes per week and severity of dry mouth, respectively) were used for illustration. Dosing strategies with two dosing categories were defined in terms of sizes of the daily doses (low and high dose) and the proportion of patients that can be expected to be preferentially treated at the low dose level. Utility and responder definitions were varied to investigate the influence on the resulting dosing strategy. By minimizing a risk function, describing the seriousness of deviations from the predefined target, optimal dosing strategies were estimated using mixture models in NONMEM. The estimated dose ranges for oxybutynin were similar to those recommended. The optimal individualization conditions were dependent on the definitions of responder and utility. The predicted gain of individualization given utility and responder definitions used was greater, when a responder criteria was maximized compared with maximizing utility.