Coal-Mine Canaries in Clinical Psychology: Getting Better at Identifying Early Signals of Treatment Nonresponse

Sequential, multiple-assignment, randomized trials (SMARTs) have emerged as a preferred design strategy with which to inform dynamic mental-health treatment decisions and adaptive interventions, yet their potential to improve patient outcomes is only as strong as the extent to which selected tailoring variables (i.e., interim response factors that dictate whether treatment shifts course) do indeed predict ultimate response. To date, tailoring variable selection has rarely drawn on adequately powered findings or conceptual links to interim target mechanisms underlying treatment response. Building on early work in this area, we detail a strategy that leverages randomized controlled trial data to simultaneously compare candidate tailoring variables at candidate decision points and their relationships with treatment response. Findings from such efforts can improve the conceptual clarity and efficiency of SMARTs, laying a foundation for modern clinical trials to ask, "Are treatment-related change mechanisms being affected and, if not, what is the most appropriate next treatment strategy?"