Distinguishing between clinical interventions and implementation strategies in trials

Abstract

Abstract Background Trialists sometimes find it difficult to determine whether the innovation they are evaluating is a clinical intervention or an implementation strategy. The inaccurate classification of innovations can lead to negative consequences such as the misspecification of the trial’s primary outcome, inappropriate choice of comparators, and suboptimal decisions about site-selection, participant eligibility, masking and fidelity assessment/control. Likewise, inaccurate classification can lead to the mislabeling of the type of hybrid effectiveness-implementation trial, the incorrect choice of reporting frameworks, and/or the inappropriate inclusion or exclusion of trial findings from systematic reviews and meta-analyses. In this paper we propose a systematic process for classifying an innovation as a clinical intervention or an implementation strategy. Main body We propose that logic models be used to provide structure to the classification process. Specifically, causal pathway diagrams can be used to identify the innovation’s mechanisms of action, treatment effect modifiers, proximal outcomes, and distal outcomes that ultimately clarify whether the innovation is a clinical intervention or an implementation strategy. We propose that innovations primarily be classified by determining the type of mechanism(s) of action through which it operates. In cases where it is unclear whether the mechanism of action is clinical or implementation in nature, we propose determining whether its proximal outcome (i.e., the most immediate outcome that is realized because of the mechanism of action) is a clinical outcome as defined by NIH and the FDA or an implementation outcome as defined by the RE-AIM, CFIR, and/or Proctor implementation outcomes frameworks. We also propose that this process may lead to some innovations being classified as having both clinical and implementation components. The proposed classification process has benefits over alternative approaches, including basing the decision on the actor/level targeted by the innovation, who delivers the innovation, whether the innovation itself needs an implementation strategy to promote its successful deployment, or whether the innovation affects clinical outcomes. Conclusions In cases where different trialists using this classification process come to different conclusions, we still believe there is value in justifying the logic of the conclusion.