Our idea with the phases of research was just that, in our view, there needs to be an order to the research and development of new intervention techniques. That there was, in fact, very little order in the studies of intervention of young children. So, we were hoping to make that point. We said, well, what phases or stages of development are already out there. The obvious types of stages or phases were what you used in drug research or the development of a surgical device or something of that sort.
There are generally four or five stages, it depends on how they are broken out. The problem in applying those to children’s speech and language disorders is the first two phases deal with things like the safety, and the dosages which are safe versus non-safe — not necessarily the ones that work versus the ones that don’t work. That is, trying to identify the highest dosage an individual can get without having safety issues. I didn’t see any particular correlate to that in children’s language. Because even though I believe it’s possible to do language intervention in a way that could actually be harmful, it could actually slow the child down as opposed to speeding them up and facilitating their progress, I don’t see anything that’s quite like a Phase 1 study that would deal with safety or even Phase 2 still dealing with safety and dosage, and to a lesser extent the efficacy.
So the question was, what is out there? If you look at the literature, we chose the use of sentence recasts. We chose it just because there are a number of studies out there, and we thought we might be able to use those studies as examples of what might reflect useful stages of development.
Phase 1: Preclinical Studies
So we started with preclinical studies. That is where the investigator may or may not be thinking about intervention with what they’re examining. It may involve correlational studies, for example, where they try to relate certain variables today or earlier with later language progress. Those sorts of studies, with a correlational design would be preclinical.
The beauty of preclinical studies, for an interventionist, is that oftentimes the idea for an intervention technique could actually come from that. A correlational study can’t determine cause and effect relationships. But it can give us ideas and allow us to generate hypotheses about what might, in fact, have a cause and effect relationship that would be a positive influence on development for children.
Phase 2: Feasibility Studies
The preclinical studies lead then to what Liza and I — I didn’t credit Liza Finestack, she’s my former PhD student and a person who’s been involved with me and wrote a chapter in a book about these phases. Anyway, in a feasibility study, those should be studies where now, at this point, the investigator is intentionally doing something that they believe will have a positive impact on whatever aspect of language it is that they are dealing with, but they are not taking all the steps necessary, they’re not sufficiently controlling the study and ruling out the subjectivity of the study to determine cause and effect.
Feasibility studies are studies that are done largely to ask questions like: Will children tolerate this? Will they play this or do this sort of activity, or is it so boring or so tedious or somehow onerous that a young child simply won’t attend and do the activity.
Feasibility studies may have some interest, and usually do have interest in effects of those things that are being tried out. But again, they’re not sufficiently experimental to allow the investigator to make a claim that the intervention technique that is being used is having a positive effect. So you get pre/post types of designs. No control group types of designs. Which are fine for showing that as long as you reinforce them for their participation children will do this task for a long time. Or they’ll only do it for five minutes at a time, and you have to then do something else. Those are the types of questions that we’re asking with feasibility. We’re taking a hypothesis that may have been developed at the preclinical level, and now we’re actually trying the technique, but we’re not yet doing the experiment.
Phase 3: Early Efficacy Studies
The experimental levels come next. The first level there (Phase 3) we call the early efficacy level. The second level then after that, which is the fourth in our series, I refer to those as later efficacy studies.
In efficacy studies, now the investigator is interested in making a demonstration that the independent variable — that is, the intervention, whatever it is — has or does not have a positive impact on the outcome. In order to really control this tightly, however, because the investigator is really interested in the internal validity of the study, they may choose an outcome variable that is not highly functional. It’s something that is very measurable. It gives a good indication that progress has been made, so you can see a change from the pre to the post.
An early efficacy study would have experimental control, ideally it would have randomization of subjects, where the investigator randomly assigns children to either receive the intervention or to receive something else or to receive nothing from the project. Generally with kids in these sorts of situations, you can’t withhold intervention that they are already receiving, so those types of interventions usually go on in the background. But the idea in an early efficacy study is to answer the question: Is there a relationship between the intervention that I got way back here by attending to a study on normal language development by someone who had no interest in, necessarily, language-impaired children, applied it in a feasibility study to determine this is a package of intervention techniques that kids will seem to enjoy and that seems to have some potential for being efficacious. Now we’re actually testing that hypothesis — that this intervention is efficacious.
The problem is the way we measure the outcome is not sufficiently functional. In all likelihood, it will be some sort of probe that is developed and done in a therapy-like situation. That’s not bad from an experimental point of view, because we can control those variables, and we can choose outcome variables that are measurable and reliable and sensitive to change or sensitive to development. So they change over a developmental period.
Phase 4: Later Efficacy Studies
The later efficacy studies differ from the early efficacy studies largely in terms of the interest in hearing out the study in more lifelike situations, and primarily measuring the outcomes in situations that are really different from the context in which the intervention actually takes place.
We may intervene on half a dozen very specific grammatical forms, for example, as part of the intervention. But we measure the outcome using some omnibus measure like mean length of utterance or developmental sentence scoring where you generate a language sample as opposed to doing a probe that’s tied to the intervention task. You may simply play with the child or get them to respond to a book or something like that. Something that does not look like the intervention, in a context that’s different from the intervention, using a measure that is more functional. That would distinguish the early efficacy study from the later efficacy study.
Phase 5: Effectiveness Studies
The final stage is actually the final stage in the drug sequence as well. That’s what we call effectiveness. We make a distinction between efficacy and effectiveness, even though the dictionary looks at them as roughly synonymous. In healthcare, they are definitely not. Efficacy studies are done under laboratory conditions, where the focus is really on the internal validity of the study. We’re trying, more than anything else, to show a clear relationship between the use of the intervention and positive outcomes. We modify the conditions of the study as much as necessary so that we won’t miss that effect if it’s really there.
In an effectiveness study, now we’re really trying to say, what would the outcome of this particular intervention be if we allowed people to administer it in really lifelike, non-laboratory types of contexts.
For example, let’s say somebody proposes that there’s yet another thing that aspirin does. The first trials would be done under very carefully controlled conditions, where there is very close monitoring of the dose, making sure patients take it and they take it on time and so forth and so on. But that’s not like the real world. In the real world, people get the prescription and they get the dosage and they may take the pills one day and forget the next two days, then make up for those two days. There are all kinds of things that can happen in the real world that are different from that controlled laboratory situation.
So the final phase would be evaluating the treatment under more lifelike circumstances, with outcomes that are really broad and highly functional — the types of outcome measures like whether clinicians actually use, when they’re working with families, the interventions designed to really have an effect on the way children communicate. That would be the highest level, and we have very few studies like that.
As a matter of fact, what usually happens is a technique will be studied up to what we call the early efficacy phase, and then there may or may not be anything that goes from that at all — but there are at least two potential phases that often times get skipped.
There’s nothing magical about our five stages. They may turn out to be not the best stages, or the way we’ve defined the boundaries between those may turn out to be not the best ones. But if investigators try to go all the way to early efficacy or later efficacy too soon, then they are likely to waste a lot of time or a lot of money. They would be much better off focusing on preclinical or feasibility studies, making sure those issues related to the feasibility of their approach, identifying outcome measures that are reliable and sensitive to change, we need to invest a lot in that type of study because we have less to lose by doing that. If we can demonstrate that a study is very feasible, then it becomes much easier to ask for more money to do the early and later efficacy studies and eventually effectiveness.
So again, the basic idea is: Can we get a sequence of developmental phases of research, where the earlier ones are relatively less risky, less costly, and yet answer questions that we really need to have answered about the intervention. And then get more funding that’s necessary to do the other studies.
The other thing that’s crucial that I should point out is that I think it’s absolutely crucial that there be some sort of depository for studies that I’m referring to as feasibility studies. If there’s not a place where people can put those studies so everyone can benefit and gain knowledge about things like the best outcome measures and the child’s tolerance and how happy they are with the interventions and so forth. If people are going to bother to take the time to get that sort of information, there has to be a place for them to put it, or they are going to be committing academic suicide, because these studies take so long and so much energy that if an investigator can’t study these studies — which many people call pilot studies — then I think we’re in for a lot of trouble, and it’s not likely that we’re going to get the efficacy and effectiveness studies that we really would hope to get.
On the other hand, we might be able to enlist more individuals in becoming intervention scientists if they have these codified phases or steps that they are expected to go through, and if there are journals that understand that part of their function is to publish well-described, well-motivated, and theoretically based feasibility studies of interventions that would seem, from a theoretical perspective, to have some merit.
Fey, M.E., & Finestack, L.H. (2009). Research and development in child language intervention: A five-phase model. In Schwartz, R. G. (Ed.), Handbook of child language disorders (pp. 513–529). New York: Psychology Press.
Rogers, M.A. (2009, June). What are the phases of intervention research? Access Academics and Research. Available at http://www.asha.org/academic/questions/PhasesClinicalResearch/