Cluster-Level Dynamic Treatment Regimens and Sequential, Multiple Assignment, Randomized Trials: Estimation and Sample-Size Considerations

• Activity Number: 262
• Type: Contributed
• Date/Time: Monday, August 1, 2016 : 2:00 PM to 3:50 PM
• Sponsor: Biometrics Section
• Author(s): Timothy NeCamp* and Amy Kilbourne and Daniel Almirall
• Companies: University of Michigan and University of Michigan Medical School and University of Michigan Survey Research Center
• Keywords: adaptive interventions ; sequential multiple-assignment randomized trial (SMART) ; dynamic treatment regimes

Abstract

Individual-level dynamic treatment regimens (DTRs), also known as adaptive interventions, are used to sequence treatment decisions based on a person's changing health information. We introduce cluster-level DTRs where treatment decisions are made at the cluster level. We present cluster-level sequential, multiple-assignment, randomized trial (SMARTs) designs, designs used to develop high-quality DTRs, in which randomization is at the cluster level and outcomes are at the individual-level. We develop a weighted least squares regression approach to compare treatment regimes embedded in a cluster-randomized SMART. We also develop a sample size calculator for designing cluster-randomized SMARTs. The validity and robustness of the calculator under various settings, including when working assumptions are violated, are evaluated through simulation. To illustrate our methods we utilize Adaptive Implementation of Effective Programs Trial (ADEPT), a cluster randomized SMART currently being conducted.