Doctoral Candidate Liwen Wu defends her dissertation on "Interim Monitoring and Sample Size Adjustment in Sequential Multiple Assignment Randomized Trials."

Advisor: Abdus Wahed, PhD, Department of Biostatistics

Committee Members: 

• Yu Cheng, PhD, Department of Statistics
• Ying Ding, PhD, Department of Biostatistics
• Chaeryon Kang, PhD, Department of Biostatistics

ABSTRACT:

A sequential multiple assignment randomized trial (SMART) facilitates comparison of multiple adaptive treatment strategies (ATSs) simultaneously. Such design becomes increasingly popular in the management of chronic diseases, such as mental health disorders. Previous studies have established a framework to test the homogeneity of multiple ATSs by a global Wald test through inverse probability weighting. However, SMARTs are generally more resource-intensive than classical clinical trials due to the sequential nature of treatment randomization in multiple stages. Thus, it would be beneficial to add interim analyses allowing for early stop if overwhelming efficacy is observed. In the first part of this dissertation, we introduce group sequential methods to SMARTs to facilitate interim monitoring based on multivariate chi-square distribution. Simulation study demonstrates that the proposed interim monitoring in SMART (IM-SMART) maintains the desired type I error and power with reduced expected sample size compared to the classical SMART. Lastly, we illustrate our method by reanalyzing a SMART assessing the effects of cognitive behavioral and physical therapies in patients with knee osteoarthritis and comorbid subsyndromal depressive symptoms.

Clinical trials are often designed based on limited information about effect sizes and variances, especially for SMARTs where strategy effects are based on sequences of treatments. Sample size re-estimation adds flexibility in adjusting sample size during the trial to ensure adequate power. Although this adaptation is popular in standard clinical trials, no method is available to perform sample size re-estimation for SMARTs. In the second part of the dissertation, we extend our knowledge about the multivariate chi-square distribution and propose a sample size re-estimation procedure for SMARTs. Sample sizes are re-calculated at interim analysis based on conditional power derived from bi-variate non-central chi-square distribution. We demonstrate through simulation study that even with an underpowered initial sample size due to misspecified parameters, the proposed method can maintain desirable power, and additional resources are only invested in trials that show promising conditional power at interim analysis.

Public health significance: SMART trials are popular in testing adaptive treatment strategies in mental and behavioral health research. Our proposed methods will increase the efficiency of testing treatment strategies for depression, addiction, and many other chronic diseases that rely on continued treatment. This research is the first of its kind to allow interim monitoring based on a global test of treatment effects in the SMART setting. The proposed designs are easy to implement and enhances the practical efficiency of the current SMART design. We believe this work will facilitate designing more SMARTs in the future.