The efficacy-effectiveness distinction in trials of alcohol brief intervention

Three recent sets of null findings from trials of alcohol brief intervention (BI) have been disappointing to those who wish to see a reduction in alcohol-related harm through the widespread dissemination of BI. Saitz (7) has suggested that these null findings result from a failure to translate the effects of BI seen in efficacy trials, which are thought to contribute mainly to the beneficial effects of BI shown in meta-analyses, to effectiveness trials conducted in real-world clinical practice. The present article aims to: (i) clarify the meaning of the terms "efficacy" and "effectiveness" and other related concepts; (ii) review the method and findings on efficacy-effectiveness measurement in the 2007 Cochrane Review by Kaner and colleagues; and (iii) make suggestions for further research in this area. Conclusions are: 1) to avoid further confusion, terms such as "efficacy trial", "effectiveness trial", "clinical representativeness", etc. should be clearly defined and carefully used; 2) applications of BI to novel settings should begin with foundational research and developmental studies, followed by efficacy trials, and political pressures for quick results from premature effectiveness trials should be resisted; 3) clear criteria are available in the literature to guide progress from efficacy research, through effectiveness research, to dissemination in practice; 4) to properly interpret null findings from effectiveness studies, it is necessary to ensure that interventions are delivered as intended; 5) in future meta-analyses of alcohol BI trials, more attention should be paid to the development and application of a psychometrically robust scale to measure efficacy-effectiveness or clinical representativeness; 6) the null findings under consideration cannot be firmly attributed to a failure to translate effects from efficacy trials to real-world practice, because it is possible that the majority of trials included in meta-analyses on which the evidence for the beneficial effects of alcohol BI was based tended to be effectiveness rather than efficacy trials; and 7) a hypothesis to explain the null findings in question is that they are due to lack of fidelity in the implementation of BI in large, organizationally complex, cluster randomized trials.