Home Research Areas The Multiphase Optimization Strategy The Sequence of Steps in MOST

We recommend that investigators who wish to use MOST to build and evaluate an optimized intervention follow this sequence of steps. Steps a – e are aimed at building an optimized intervention. Step f then evaluates the optimized intervention to confirm that it produces a statistically better outcome than a suitable control or comparison group.

1. Establishment of theoretical models of (i) the health behavior to be intervened on and (ii) the intervention process. These models will be informed by theory, scientific literature, clinical experience, exploratory data analyses, and other sources of information.
2. Identification of a set of candidate intervention components to be examined. These components may be parts of the intervention itself (e.g. parts of the curriculum in a school-based intervention), aspects of intervention delivery (e.g. telephone vs. in-person counseling), and/or factors that impact compliance/adherence/participation (e.g. providing child care). Any pilot studies to investigate the feasibility of intervention components are done in this step.
3. Screening of candidate intervention components. Here the term “screening” refers to taking the set of candidate intervention components identified in step b and selecting the ones that appear to be having the desired impact on intervention outcomes. In other words, in this step the idea is to identify the active intervention components out of the list of candidate intervention components. This is done by means of randomized experiments. Care is taken to choose efficient experimental designs for these experiments, and to size them appropriately.
4. Refining of intervention components. In this step, the components that have been selected in step c may be further refined with the objective of arriving at optimal levels or doses. For example, if one or more components may take on many levels (e.g. number of hours of counseling), further experimentation may be performed to identify the level that gives the best outcome. (The refining step is not always needed.)
5. Assembly of draft intervention. At this point, the investigator has identified a set of active intervention components and their optimal levels/doses. In the assembly step, these components are put together into a draft intervention. Based on the experiments performed in the screening and refining steps, the investigator can form a rough estimate of the expected size of the effect of the intervention. If based on this the intervention is deemed potentially efficacious, effective and/or cost-effective, the investigator proceeds to the next step.
6. Confirming the efficacy/effectiveness/cost-effectiveness of the intervention. This is a standard intervention evaluation done via a two-group RCT. Note that the RCT can be sized with considerable precision based on the information gathered in the previous steps of MOST.

Our perspective is that optimization is an ongoing process. Thus while the optimized intervention is being implemented, the investigators may wish to conduct MOST again to continue to improve the intervention.