By Kari Kugler, Ph.D.
What do smoking cessation, weight management, cancer, drug abuse, and related conditions such as HIV/AIDS and hepatitis C have in common? They have all been positively impacted by behavioral interventions for prevention and treatment. However, scientists face a major conundrum: how best to optimize these interventions in order to achieve the greatest public health benefits.
The Office of Behavioral and Social Sciences Research and the National Institute on Drug Abuse are sponsoring a training, Optimization of Behavioral and Biobehavioral Interventions, to explore in depth some of the answers to this question.
Linda M. Collins, The Pennsylvania State University; Susan Murphy, University of Michigan; and Daniel Almirall, University of Michigan will lead the training May 16–20 in Bethesda, Maryland. Applications for this 5-day event are now being accepted.
Behavioral and biobehavioral scientists interested in using innovative study designs to uncover how and for whom interventions work, and who wish to develop highly effective, efficient, economical, and scalable interventions, will benefit from this training. Drs. Collins, Murphy, and Almirall will help attendees learn about emerging alternatives to the classical approach to developing behavioral and biobehavioral interventions. These new approaches are aimed at optimizing behavioral and biobehavioral interventions.
Not the Typical Behavioral Intervention Training
The classical approach to developing multicomponent behavioral and biobehavioral interventions has been to identify a set of intervention components, assemble them into a package, and evaluate the package by means of a randomized control trial (RCT). While acknowledging the RCT’s undisputed role in intervention evaluation, some behavioral scientists (Collins, Kugler, & Gwadz, 2016) have critiqued reliance on the RCT for intervention development.
Based on the results of an RCT alone, it is not possible to determine which intervention components contribute to any observed treatment effect, which are inactive, and which may have negative effects. In other words, most behavioral and biobehavioral interventions are essentially black boxes. This is preventing the field of behavioral science from developing a coherent base of knowledge to build on and is obscuring the way forward to improve behavioral and biobehavioral interventions.
A new paradigm is emerging, inspired by procedures that are used widely in engineering and systems science. In this new paradigm, called the multiphase optimization strategy (MOST), behavioral interventions are optimized to meet a specific criterion before they are evaluated. For example, optimization might be aimed at many things, including:
- Weeding out inactive or counterproductive components,
- Developing an intervention that meets a particular defined standard of effectiveness,
- Engineering the most cost-effective intervention, and
- Selecting the best set of tailoring variables and decision rules in a time-varying adaptive intervention.
Optimization requires conducting experiments, usually using approaches other than the RCT. The most appropriate experimental design depends on the intervention and the goals of the optimization. However, it is always aimed at gathering empirical information on the individual and combined performance of the components being considered for inclusion.
Intervention Training Topics
This May, the Training on Optimization of Behavioral and Biobehavioral Interventions will discuss MOST for both fixed (traditional) and interventions. Adaptive interventions, also known as "adaptive treatment strategies" or “dynamic treatment regimens”, are individually tailored treatments in which a sequence of decision rules specifies how the intensity or type of treatment should change to reflect the patient's needs. Interventions that adapt improve participant outcomes, while decreasing the cost and burden of the intervention. One special type of adaptive intervention is the just-in-time adaptive intervention (JITAI), in which real-time information on the participant is used to inform the immediate delivery of intervention options.
A variety of experimental designs utilized to gather information and optimize an intervention for the MOST framework will be discussed. Some examples include factorial and fractional factorial screening experiments; the sequential, multiple assignment, randomized trial (SMART; Almirall, Nahum-Shani, Sherwood, & Murphy, 2014) for optimization of adaptive interventions; and microrandomized trials for optimization of JITAIs (Klasnja, Hekler, Shiffman, Boruvka, Almirall, & Tewari, 2015).
In addition, there will be presentations from behavioral and biobehavioral scientists who have experience implementing all of these approaches. There will also be a panel discussion in which NIH program officers will offer advice about funding.
The group size will be limited, and there will be ample time for discussion and interaction. Attendees will be encouraged to raise questions and make comments about relevance to their own work, and there will be opportunities to arrange one-on-one meetings with instructors and presenters.
For more information about the training and how to apply to attend, please visit:
Almirall, D., Nahum-Shani, I., Sherwood, N. E., & Murphy, S. A. (2014). Introduction to SMART designs for the development of adaptive interventions: with application to weight loss research. Translational behavioral medicine, 4(3), 260-274.
Collins, L. M., Kugler, K. C., & Gwadz, M. V. (2016). Optimization of multicomponent behavioral and biobehavioral interventions for the prevention and treatment of HIV/AIDS. AIDS and Behavior, 20, S1-S18. doi: 10.1007/s10461-015-1179-7.
Klasnja, P., Hekler, E. B., Shiffman, S., Boruvka, A., Almirall, D., Tewari, A., & Murphy, S. A. (2015). Microrandomized trials: An experimental design for developing just-in-time adaptive interventions. Health Psychology, 34(S), 1220.
Kari Kugler, Ph.D. Research Associate, The Methodology Center, The Pennsylvania State University. My primary area of interest includes the development of effective and efficient behavioral interventions targeting a wide range of health behaviors among various populations and contexts. In particular, I am interested in developing interventions targeting the intersection of sexual and reproductive health, including HIV, alcohol use, and obesity among adolescents and young adults. I collaborate with Linda Collins, Lori Francis, Jennifer Savage-Williams, and colleagues at The University of North Carolina at Greensboro on projects using the multiphase optimization strategy (MOST). In addition, I work closely with Donna Coffman on using modern causal inference methods and Jennie Noll on understanding the impact of early child maltreatment on later sexual risk behaviors.