Holder, Harold D.
; Flay, Brian; Howard, Jan; Voas, Robert
; and Grossman, Michael Alcoholism: Clinical and Experimental Research
, vol. 23, issue 1, pgs. 183-194 (1999)We build on precedents from other health research to present a phases model of research for alcohol problem prevention that accommodates the special characteristics of this research. We propose a five-level model, in which research moves along a series of relevant continua: from basic to more and more applied research; from descriptive hypothesis-generating pilot studies to full-fledged, methodologically sophisticated, hypothesis-testing studies; from smaller to larger samples for testing; from greater to lesser control of experimental conditions; from more artificial "laboratory" environments to real-world geographically defined communities; from testing the effects of single prevention strategies to more complex studies of multiple strategies integrated into intervention systems; and from research-driven outcome studies to "demonstration" projects that evaluate the capacity of various types of communities to implement prevention programs based on prior evaluations. The five phases of research are: (1) foundational research to define and determine the prevalence of specific alcohol-involved problems, establish causal factors and processes that yield the specific problems or increase the risk of a problem, and provide the foundations for the development of effective prevention interventions; (2) developmental (preliminary effectiveness) studies to develop and test the likely effectiveness, safety, and costs of new interventions or to assess the effectiveness, safety, and costs of an existing intervention; (3) efficacy studies to determine the effects, safety, and costs of an intervention under optimal conditions of implementation (or availability or enforcement) and acceptance (or adoption at the community, organizational, or group level; or participation, compliance, or adherence at the individual level); (4) effectiveness studies of the real-world effectiveness of preventive interventions with purposeful or natural variation in implementation and acceptance; and (5) demonstration studies of the effects of interventions when widely disseminated. The proposed phases model for alcohol problem prevention research presented herein differs in significant ways from the models established by other National Institutes of Health agencies. Greater emphasis is placed on natural experiments, on methods development along the whole research continuum, on collapsing or combining research phases when appropriate, on recognizing the critical importance of behavioral parameters early as well as late in the research sequence, and on extending the research continuum to embrace diffusion and dissemination (i.e., technology transfer) studies. We also include examples of phased research in existing alcohol studies and a discussion of relevant issues, including cost, special populations, methods, and dissemination. If systematically followed, this model has the potential to contribute to wider testing and dissemination of prevention interventions of known effectiveness.