Twenty years of intervention optimization

Ann Behav Med. 2024 Dec 7:kaae076. doi: 10.1093/abm/kaae076. Online ahead of print.

Abstract

In the classical paradigm for intervention research, the components that are to make up an intervention are identified, pilot tested, and then immediately assembled into a treatment package and subjected to an evaluation randomized controlled trial (RCT) to assess the performance of the entire package. Intervention optimization, which adapts ideas from technological fields to intervention science in order to hasten scientific progress, is an alternative to the classical paradigm. The first article introducing intervention optimization via the multiphase optimization strategy (MOST) was published in Annals of Behavioral Medicine in 2005. In this commentary, I reflect on the evolution of intervention optimization from that first publication to today, and on what the future could hold if the intervention science field continues to adopt the optimization paradigm. I propose that if intervention optimization became standard operating procedure, the field would accumulate a coherent base of knowledge about what specific intervention strategies work, for whom, under which circumstances, and why; every intervention produced would contain only components that contribute enough to justify their resource requirements; interventions would be readily implementable; and as the knowledge base grew, interventions would be improved continually.

Keywords: intervention optimization; multiphase optimization strategy; optimization randomized clinical trial.

Plain language summary

The dominant way of developing evidence-based behavioral and biobehavioral interventions has been to construct the intervention a priori, and then determine via a standard evaluation randomized controlled trial (RCT) whether the intervention as a package has a statistically detectable effect. This approach makes it impossible to assess the performance of individual intervention components. Consequently, when an intervention shows a detectable effect, it is unknown which components were contributing to the effect and whether they are all necessary. Moreover, when an intervention does not show a detectable effect, it is unknown what went wrong and what the next steps should be. Other fields that have made considerable scientific progress in recent decades, such as technological fields, do not operate this way. This commentary discusses a new approach to development of evidence-based behavioral and biobehavioral interventions, namely intervention optimization via the multiphase optimization strategy (MOST). Based on principles drawn from fields such as engineering, economics, decision science, and implementation science, MOST will enable intervention science to hasten its progress. This commentary suggests that if intervention optimization became standard operating procedure, interventions would become more economical and implementable; the scientific knowledge base would expand; and the public health impact of interventions would increase.