Researchers from the d3center hosted a symposium at the 43rd Annual Meeting & Scientific Sessions of the Society of Behavioral Medicine in Baltimore on April 8th. During the session, titled SMART Studies for Developing Adaptive Interventions in Substance Use and HIV, co-director Daniel Almirall led a discussion about optimizing adaptive interventions with Sequential, Multiple Assignment, Randomized Trials (SMARTs). To illustrate the application of SMART experimental designs, presenters from Dartmouth College and Florida State University detailed how they each implemented SMARTs in their research. Each presenter demonstrated unique capabilities of the SMART and emphasized exciting ways SMARTs produce answers to the nuanced scientific questions that inform high-quality adaptive interventions.
Dr. Catherine Stanger, who designed and conducted a SMART to develop an adaptive intervention for adolescent cannabis users, investigated two critical questions: (1) whether adding working memory training to contingency management (monetary or other motivational incentives) would increase cannabis abstinence among study participants; and (2) whether the intensification of incentives would increase abstinence among early non-responders. In addition, she was able to see if first stage working memory training increased second stage response to intensification. Stanger’s study emphasized the capacity of SMARTs to illuminate the systemic effect produced by the introduction of new intervention options and the inter-stage modulation of established options.
In a second study that examined whether contingency management was effective at driving viral suppression among adolescents with HIV, Dr. Sylvie Naar exemplified how the SMART can be used to optimize adaptive interventions along dimensions other than effectiveness—like cost-effectiveness. Would incentives prove equally or nearly as effective if they were only offered to those who failed to respond to first line support through either cell phone support or text message reminders? At the same time, the study looked for potential synergies between incentives and phone-based support. Would incentives, added to cell phone support, make cell phone support more effective than text message reminders? This line of inquiry showed how multiple randomizations within a SMART let researchers see what intervention option components, when added to an existing intervention option, may give that option an edge on another option once shown to be comparable.
As discussant, Dr. Almirall fielded questions from an audience of 40 attendees, many of whom took pictures of slides displaying the SMART schematic. Dr. Almirall spoke to important differences between the SMART and other trial designs, some of which are useful for evaluating adaptive interventions but aren’t equipped to optimize them. In particular, he emphasized the way SMARTs equip researchers to analyze interactions between intervention options across multiple stages of intervention. One example of this are “delayed effects,” where an early intervention option doesn’t impact patient outcomes until a later stage, or until the effect is activated by a subsequent intervention option. A spirited debate about the methodology behind the SMART offered Dr. Almirall an occasion to substantiate the true novelty and versatility of the SMART.
Ultimately, the symposium provided a platform for SUD/HIV scientists to showcase how novel experimental trial designs are being used to develop the next generation of adaptive interventions for combating substance use disorder and HIV.