In recent years, research has been conducted to develop Sequential, Multiple Assignment, Randomized Trial (SMART) designs. These experimental designs were created to aid in the construction of adaptive treatment strategies for individuals, particularly in medical contexts. Simultaneously, research has been done on developing the use of randomized trials to evaluate static treatments in clustered settings. For example, methodology for evaluating academic interventions in schools has been researched extensively. The benefits individuals experience by following adaptive treatment strategies can be expected to extend to individuals in clustered settings, as well. This work helps bridge these two lines of research by proposing and evaluating experimental designs for developing adaptive treatment strategies in clustered settings. This research first proposes experimental designs in which the unit of treatment changes over the course of the trial, and in which individuals within a clustered setting are regrouped mid-trial to receive further treatment based on initial treatment response. Sample size formulae are derived to allow researchers to properly power these trials. Sample sizes for the proposed designs are evaluated in a sensitivity analysis. Next, methods of analyzing data from individual SMARTs are adapted for use with the data that would be obtained from the proposed trial designs. These adapted methods are evaluated through a simulation study. Finally, a web application is created to allow easy dissemination of the research presented here. The findings of this research suggest that complex clustered SMART designs can be feasibly implemented, presenting the opportunity to bring improved treatment strategies at lower resource requirements, as compared to existing options, to clustered settings. Furthermore, findings of this research suggest that existing SMART data analysis methods can be successfully adapted to accommodate the clustered data arising from the proposed designs.

Last modified

• 10/28/2018

Creator

• Rachel Ktsanes

DOI

• https://doi.org/10.21985/N26160

Subject

• Statistics

Keyword

• Statistics

Date created

• 2017-01-01

Resource type

• Dissertation

Rights statement

• In Copyright