127-31: Using the SAS® System for Experimental Designs for Multi-Component Interventions in Medicine

We demonstrate how to use SAS to design experiments for multicomponent interventions for multifactorial health syndromes. Multifactorial syndromes are health conditions that have more than one risk factor related to the outcome and require interventions with several components that target different risk factors. The design and analysis of multicomponent trials are complicated by the number of factors to be studied and the interdependency among the factors. A full factorial design is appropriate when there are a few risk factors to treat; however, when there are many risk factors to treat, fractional factorial designs allow for estimation of the main effects at the sacrifice of higher-order interactions (available in ADX Interface SAS/QC software). When selecting risk factors to treat, researchers must consider the inter-relationships among factors, particularly interactions (using ADX). Randomization to treatment arms needs to consider whether participants can be individually randomized (the PLAN procedure) or if the trial design requires the use of cluster randomization (the SURVEYSELECT procedure). Clinical investigators have been reluctant to design and test multicomponent interventions, both because of their greater complexity and of the concern that it is not possible to disentangle the effects of the individual components to determine those that are beneficial. Because of this reluctance many potentially effective multicomponent intervention strategies have been left untested. Furthermore, each component of an intervention adds to its overall cost and complexity; thus, designs that allow estimation of component effects could greatly enhance efficiency. The SAS System has a strong set of tools to help design multifactorial trials.