Description:

Design and Analysis of Sequential Multiple Assignment Randomized Trials (SMARTs)

When providing care for patients with chronic conditions, it is common to adapt a patient’s treatment based on his/her response to earlier treatments. In a SMART design, these interim treatment adaptations are determined by randomization for each patient, thereby allowing for unbiased evaluation of the optimal treatment option at each decision point. However, the use of SMART in practice can be limited due to the costs and logistical difficulties associated with the design. In this talk, I will give an overview of SMART designs, highlight some recent findings regarding the analysis and design of a SMART (e.g. sample size determination), and discuss its utility in quality improvement of a treatment program. The goal of our work is to enhance the practicality of SMART-type designs in chronic care management.