Xiaoyan Yin is a doctoral candidate in the Department of Department of Biostatistics and Bioinformatics, George Washington University, and works as a biostatistician in the Diabetes Prevention Program Outcomes Study (DPPOS) in Biostatistic center. Her research interests are Sequential Multiple-Assignment Randomized Trial COMparing Personalized Antibiotic StrategieS (SMART COMPASS) design, multiple comparisons, group sequential designs and infectious disease. She had experience in competitive programming, market investigation and analysis, data mining and won several awards. She served as the Co-President for the GWU Student Chapter of the ASA and received a Recognition Award for Outstanding Dedication and Commitment to the Chapter in 2018.

Abstract: SMART-COMPASS – Design, Analyses, and Software

Patient management is dynamic, a sequence of decisions with therapeutic adjustments made over time. Adjustments are personalized, tailored to individuals as new information becomes available. Strategies allowing for such adjustments are infrequently studied.

Two major treatment decisions occur during the treatment of serious bacterial infections: empiric and definitive therapies. Empiric therapy selection is based on immediately available and often limited information upon recognition of the clinical syndrome. Definitive therapy is selected once organism identification and antibiotic susceptibility testing (AST) results are known, frequently 48-72 hours later than empiric therapy selection. COMparing Personalized Antibiotic StrategieS (COMPASS) is a trial design that compares strategies consistent with clinical practice, decision-rules that guide empiric and definitive therapy decisions. Sequential multiple assignment randomized (SMART) COMPASS allows evaluation when there are multiple definitive therapy options. Sequential randomization provides the opportunity to create new strategies, which differ with respect to definitive therapy selection, and compare them in a randomized setting. Trial participants can be re-randomized to the definitive therapy based on the tailoring criterion to determine the optimal therapeutic adjustment and overall strategy. SMART COMPASS is pragmatic, mirroring clinical treatment decision-making, and addressing the most relevant issue for treating patients: identification of the strategy that optimizes ultimate patient outcomes.

Several statistical challenges arise with SMART COMPASS including how to: estimate effects and standard errors, control trial-wise error, and calculate sample size and power, since different strategies can “share” patients. Weighting of patients is required to obtain appropriate estimates of effects and associated standard errors. Estimates of the proportions of patients that will be re-randomized at the definitive treatment stage are required for sample size calculations. Multiple testing methods are necessary to control error rates as sequential randomization implies at least three strategies. Group-sequential monitoring provides important efficiencies but requires adjustments when calculating sample size and power.

Methods to design group-sequential SMART COMPASS with continuous endpoints, binary, and rank-based endpoints are proposed. Multiple testing procedures for identification of the best strategy are evaluated and guidance on procedure selection is provided. Approaches for the calculation of the sample size are discussed. R Software including a shiny app are developed to implement study design, data analysis and visualization.

Keywords: SMART COMPASS, Strategy Comparison, Personalized medicine, Pragmatic Trial, Multiple Testing Procedure, Sample size, Sequential Power, Sequential Monitoring, Shiny APP.