In the discipline of Communication Sciences and Disorders (CSD), intervention research aimed at informing clinical practice has proceeded for nearly a century now without a clear set of expectations about what the necessary, or at least desirable, steps ought to be to develop foundational knowledge that can later support inferences of causality as well as conclusions that can be generalized to the larger population being sampled. Our field may not be alone in this predicament, but fortunately, there are scholars within CSD who are addressing our need for an organizational framework around which programmatic intervention research can be developed (Fey & Finestack, 2009; Robey, 2004; Robey & Schultz, 1998). As Robey (2004) states “an organizing structure for conducting clinical-outcome research can also bring order to the task of assessing the scientific evidence in a literature base of rehabilitation research” (p. 402), While Robey focuses on a phased approach for conducting rehabilitation research, Fey and Finestack (2009) focus on a system for developing language interventions with children. They contend that “adoption of such a system is essential to the growth of language intervention science and to the development of an evidence base suitable for guiding clinicians in making crucial treatment decisions” (p. 514). The purpose of this article is to raise awareness about the five-phase model of intervention research and to suggest that these publications be integrated into coursework addressing clinical practice research in CSD.
The five-phase model for studying the effectiveness of intervention begins with the “pre-trial” studies. In Phase I research, the intervention and its hypothesized effects are identified. Generally, a small number of participants are recruited, and initial approximations of candidacy criteria are established. The treatment protocol is worked out, as are the specific outcome measures. While these studies are often observational or correlational, estimates of effect size can be established in Phase I research, and the data can be used to calculate the sample size required to gain sufficient statistical power to test the hypotheses. However, these latter goals are more commonly achieved in the next phase.
In Phase II research, labeled by Fey and Finestack (2009) as “feasibility studies”, the clinical viability of the intervention is tested. Studies conducted at this phase are considered by Fey and Finestack to be clinical trials, even though they are primarily “exploratory and preliminary with respect to intervention outcomes” (p. 520). Robey (2004) states that Phase II studies should “determine early indications of the presence and magnitude of efficacy” as well as “refine the definition of the population, the treatment protocol and develop a manual for consistent implementation and replication” (p. 404). Case studies, discovery-oriented single-subject designs, and small group cohort control studies are all appropriate at this stage. One of the most important messages about Phase II research is that these foundational studies should not be skipped by prematurely jumping to experimental tests of efficacy, as that can waste both time and money. The need to support publications of Phase II research is also underscored, and even though the data may not yield high levels of evidence, the theoretical and/or empirical motivation for the intervention approach can serve as the primary focus. Describing the intervention and the methods used to evaluate treatment fidelity are examples of highly valuable contributions that are worthy of publishing, but claims of efficacy should not be made in relation to Phase II studies. If positive outcomes are observed, certainly those findings can be used to motivate the next level of testing (Fey & Finestack, 2009).
The purpose of Phase III research is to begin to test efficacy, and accordingly, research at this phase has been labeled by Fey and Finestack (2009) as “early efficacy” studies. Many types of designs are appropriate at this stage but should be experimental, or at least quasi-experimental, in the sense that designs must entail comparisons of treatment with no-treatment control conditions or withdrawal of treatment or other experimental conditions that permit inferences of a causal relationship between the treatment and the effect. As these studies do not require large samples and should be relatively low cost, they do not support inferences concerning how the treatment effects may generalize to the larger population. Rather, internal validity is the central concern.
The purpose of Phase IV research, labeled “later efficacy” by Fey and Finestack (2009), is to directly compare the target intervention with an alternative intervention or a no-treatment condition to address causality but under more generalizable conditions (i.e., typically with more participants). According to Fey and Finestack (2009), “later efficacy studies provide clear evidence of a target intervention’s efficacy under ideal laboratory conditions” (p. 525). Even though studies at this phase may be used to guide clinical decisions, they are not the highest level of evidence needed to support clinical decision making.
Finally, the focus of Phase V studies, or “effectiveness” research, is on determining whether “the therapeutic effect is realized in day-to-day clinical practice” (Robey, 2004, p. 405). Generally, Phase V research is considered “field research” or “community-based research” and is designed to test generalization of the intervention to a larger sample and under typical and somewhat variable clinical contexts. A central question for Phase V research is whether the effects are similar to those found in later efficacy studies (Fey & Finestack, 2009) and to determine who benefits from the treatment (Robey, 2004). It is during this phase of research that questions concerning the cost-benefit ratio of the intervention can be addressed.
The organizational framework provided by the five-phase model of intervention research is paramount for researchers, reviewers, clinicians, students, and policymakers to understand, as it highlights the important role that feasibility and early phase research play in developing the foundation for later efficacy and effectiveness research. As effectiveness research is not necessarily the ideal means to address causality (as there is less control over potential confounds, and field research is associated with greater variability as compared with efficacy studies), there is great danger in moving too quickly to large sample randomized-control field trials, as these designs, in and of themselves, are an inadequate basis upon which to infer that the observed effects actually stem from the treatment itself as opposed to some uncontrolled confound. Fey and Finestack (2009) conclude by suggesting that Phase V research may not actually be an endpoint but rather part of the process in that further hypothesis testing at Phase III and IV levels may be indicated after a Phase V study to determine, for example, which factors may distinguish between participants who maintained gains and those who did not after treatment was terminated. Additionally, the five-phase model of intervention research validates that early phase research is important, in fact necessary, and should be encouraged, more frequently published, and required by funding agencies to provide foundational support for Phase V studies.