Alternatives to Randomized Control Trial Designs for Community-Based Prevention Evaluation

David Henry, Patrick Tolan, Deborah Gorman-Smith, Michael Schoeny

Multiple factors may complicate evaluation of preventive interventions, particularly in situations where the randomized controlled trial (RCT) is impractical, culturally unacceptable, or ethically questionable, as can occur with community-based efforts focused on inner-city neighborhoods or rural American Indian/Alaska Native communities. This paper is based in the premise that all research designs, including RCTs, are constrained by the extent to which they can refute the counterfactual and by which they can meet the challenge of proving the absence of effects due to the intervention-that is, showing what is prevented. Yet, these requirements also provide benchmarks for valuing alternatives to RCTs, those that have shown abilities to estimate preventive effects and refute the counterfactual with limited bias acting in congruence with community values about implementation. In this paper, we describe a number of research designs with attending examples, including regression discontinuity, interrupted time series designs, and roll-out randomization designs. We also set forth procedures and practices that can enhance their utility. Alternative designs, when combined with such design strengths, can provide valid evaluations of community-based interventions as viable alternatives to the RCT.