Skip to content

Strategy offers a better way to evaluate health interventions

There’s a better way to evaluate the effectiveness of health interventions. It can help make health interventions not only effective, but scalable, affordable and less complex. When testing an intervention that is trying to modify health attitudes or behaviors – like a smoking cessation program – the Multiphase Optimization Strategy, or MOST, approaches the research process from a vantage point borrowed from the engineering world.

Penn State Clinical and Translational Science Institute recognized the potential of this approach and helped fund part of the continued work by principal investigator Linda Collins, PhD. MOST consultations have been given to faculty of Clinical and Translational Science Award programs across the country.

“My collaborators and I started this research because of what we see as serious problems in the behavioral and biobehavioral intervention fields,” Collins, Distinguished Professor of Human Development and Family Studies, said. “Hundreds of millions of dollars have been spent on research in this area for more than 30 years. Yet, for the most part, the field has failed to develop a coherent body of scientific evidence about what works and what doesn’t work. Many interventions developed in academic settings at great expense never go to scale and steady incremental progress is not being made.”

After discussing these issues with a colleague at University of Michigan, Susan Murphy, the idea for MOST started with a hallway conversation Murphy had with Vijay Nair, an eminent engineering statistician.

“When she explained how interventions are typically developed and evaluated, Vijay responded that an engineer would approach the problem quite differently,” Collins explained. “Susan knew I would be interested in this perspective and invited me to Michigan for a couple of days for an intensive discussion.“

That collaboration led to a first paper, “A strategy for optimizing and evaluating behavioral interventions” in Annals of Behavioral Medicine in 2005.

Examples of studies that could benefit from MOST include testing a new smoking cessation program, a school-based drug prevention program or a weight loss program. What makes MOST work for these studies is that the interventions are made up of multiple components. For example, a smoking cessation program may include an education component, a support component (like a help line) and a reward component to encourage smokers when they don’t light up a cigarette. The components form a package. Traditionally, a randomized control trial is used to evaluate the package to see if it works better as a whole compared to another intervention or a control group.

But there is a potential problem with sole reliance on evaluating an intervention this way: The sum of the parts is evaluated, not the parts individually. If the research shows that the intervention as a whole helped a person stop smoking, it may not be apparent which components contributed to the positive effect or whether one component influenced the effect of another. If the research shows that the intervention did not help, then it can’t be determined if any of the components should be kept or if the negative effects of one component canceled out the positive effects of others.

“The way interventions are typically developed today, the effectiveness of the intervention is the only consideration,” Collins said. “Implementation cost, the amount of staff or participant time required, the complexity of the intervention, and similar factors are not considered in intervention development. This means that a highly effective intervention can be developed, and ultimately be impractical because it is too costly, burdensome or complex to be scalable. MOST makes it possible to develop an intervention that is immediately scalable, because these considerations are taken into account from the beginning. “

MOST incorporates elements of both biobehavioral and engineering sciences to identify the intervention that provides the highest expected level of effectiveness obtainable within key constraints imposed by the need for efficiency, economy and scalability. It is thinking differently about how to do research.

“I’ve been selling MOST to researchers for nearly 15 years, with varying degrees of success,” Collins said. “In the beginning, many intervention scientists saw the MOST approach as extremely radical, even though it is based on ideas and approaches that have been standard operating procedure in engineering, agriculture, and many other fields for more than a century. Because of this, I think many people felt a project using MOST would not be fundable, and it would be hard to publish any results. A few brave souls were willing to take MOST out for a ‘test drive’ early on. I am so grateful to them, because being able to point to successful funded research based on MOST has made a huge difference.”

Collins continues to develop MOST, collaborating with intervention scientists on a wide variety of applications of the strategy. She has also been working with collaborators to add multi-criteria decision analysis into the MOST framework. She recently published a book, “Optimization of Behavioral, Biobehavioral, and Biomedical Interventions: The Multiphase Optimization Strategy.”

Collins will present “Bringing Health and Education Interventions into the 21st Century” at the College of Health and Human Development Schmitt Russell Lecture at 4 p.m. Nov. 7. This lecture will be held at Bennett Pierce Living Center, 110 Henderson Building at Penn State University Park. A reception will precede the lecture at 3:30 p.m.

Learn more about the MOST approach and see an in-depth explanation of how it is conducted here. Studies that have used the MOST approach include this intervention after an acute coronary syndrome and this physical activity intervention for breast cancer survivors.

If you're having trouble accessing this content, or would like it in another format, please email the Penn State College of Medicine web department.