Outlook: Newsletter of the Society of Behavorial Medicine

Fall 2019

SBM Past-President Dr. Gary Bennett Shares His Experiences Using the Multiphase Optimization Strategy (MOST)

Tiffany Bullard; Optimization of Behavioral and Biobehavioral Interventions (OBBI) SIG Student Liaison

Gary G. Bennett, PhD

The Society of Behavioral Medicine’s (SBM) Optimization of Behavioral and Biobehavioral Interventions Special Interest Group (OBBI SIG) recently interviewed SBM Past-President, Dr. Gary G. Bennett, on his experiences optimizing a standalone text messaging-based weight loss intervention as part of an ongoing R01 trial.

Tell us a bit about the aims of your grant.

We are using MOST to identify which components should be included in a standalone text messaging obesity intervention for adults. Using a full factorial design, we are randomizing 592 participants to one of 16 experimental conditions. Findings from this study will help us determine which text messaging components and levels contribute to 6-month weight change and the 6-month change in diet and physical activity. We will also assess the proportion of participants who achieve > 5% weight loss at 6 months and maintenance at 12 months.

Why was it important to use MOST to answer your specific research question(s)?

There are so many reasons! First, I love the philosophy of optimizing a treatment package for a specific purpose. I find it to be both liberating conceptually while forcing a degree of focus that’s lacking in other approaches. 

Second, iteration. What MOST describes as the continuous optimization principle could also be described as iteration, a concept that is fundamental to modern software design. MOST hasn’t been around long enough to actualize this principle, but in theory, after completing optimization, we should re-start MOST – either on our work or on that of others – to continue refining our interventions for optimal effectiveness.

Finally, the component selection phase is a critical hedge against intellectual hubris! Put another way; when developing interventions, we make so many decisions that have little to no evidentiary basis. We make them with the best intentions, extrapolating from theory and the available evidence, but these decisions are not data-driven. Also, we’re often wrong (and don’t know it).

Describe one major challenge that has emerged while conducting your study.

Just one?! Selecting components to makeup our optimized treatment package is challenging because there are so many considerations: the criteria one uses to make the selection, the outcomes of interest, their measures, and analysis, how one should think about interactions (to be clear, all the action is in the interactions), not to mention the atmospherics of the selection process (who leads the discussion, how the data are presented, how discrepancies are resolved), and in our case, and how one keeps the selection meeting(s) on track (our findings were so surprising that we kept tripping in rabbit holes). None of these challenges, however, outweigh the fun of MOST.

What advice do you have for other behavioral scientists considering the MOST framework for their research?

Linda Collins and her team at Penn State have done an amazing job of disseminating MOST! I refer to their materials frequently. We’ve sent colleagues to their training courses, and they’re very open to collaboration and advice. I also learn more rapidly by listening and I’ve loved watching their talks on YouTube, so, I’d recommend contacting them.

For background information and additional resources on optimization of behavioral and biobehavioral interventions, visit The Penn State Methodology Center.