A general framework for dose optimization.

Dose optimization is a ubiquitous challenge in clinical practice and includes both pharmacologic and non-pharmacologic interventions. Methods for the statistical assessment of optimum dosing are lacking. We developed a generic framework for dose titration and demonstrated its application in two domains. Optimum warfarin dose was estimated from clinical titration data. In addition, cardiac pacemaker interval optimization was conducted using three conventional techniques. For both data types, optima were obtained from mathematical functions fit to the raw data. The precision of the estimated optima was quantified using bootstrapping. In pacing optimization, the observed precision varied significantly among the techniques, suggesting that impedance cardiography is superior to commonly used echocardiographic methods. The average 95% confidence interval of the estimated optimum warfarin dose was +/-18%, suggesting that titration within this range is of limited utility. By identifying statistically ineffective interventions, objective analysis of optimization data may both improve outcomes and reduce healthcare costs.