How can stammering in young children be treated effectively if the causes are not known? This is the edited version of an email debate that grew out of a lively discussion between professors Mark Onslow and Scott Yaruss at the 2005 Oxford Dysfluency conference.
The diagnosogenic theory stated that stuttering was caused, in part, by parents inappropriately drawing attention to a child's otherwise normal disfluencies. In recent years, many have commented on the numerous shortcomings of this theory as an explanation for early stuttering, and, in particular, on the negative effects it has had on treatment planning and clinical decision making.
Given your recent research on stuttering treatment, combined with research on recovery, with what theoretical framework would you replace the diagnosogenic theory, and how would such a theory help to explain basic phenomena associated with childhood stuttering?
There should currently be no replacement for the diagnosogenic theory as a driver of treatment for early stuttering. Many theories are available (for an overview, see Packman & Attanasio, 2004), but none of them has proven to be correct, and only one of them - or perhaps none of them - is correct. Hence for the time being, it is a dubious practice to base treatment on any theory of what causes or perpetuates stuttering.
We are all desperate to find out what causes stuttering. But intervention of early stuttering can occur independent of efforts to uncover its cause. For example, the Lidcombe Program is not driven by a theory of the cause of stuttering, and is nonetheless efficacious according to a recently published randomised controlled trial (Jones et al., 2005).
Few would doubt the efficacy of the Lidcombe program, as described in numerous publications and the recent clinical trials. Still, some may question its effectiveness in everyday clinical settings. Further, is Lidcombe the only way to accomplish our common goal of eliminating stuttering in young children?
Of course, theory has not always served our field well. Many clinicians still cling to ineffective treatments derived from the long-disproved diagnosogenic theory, but would this necessarily have to be the case with other theories?
Improvements in treatment might still be achieved through the rigorous application of theory-driven clinical research aimed at uncovering factors involved in the onset and development of the disorder. Theories that focus on single factors have proven unsatisfactory throughout the history of our field, so I would start with a multi-factorial theory that incorporates more than one factor as a potential cause for stuttering.
I hope that no clinician is still using the diagnosogenic theory to treat stuttering. That would be disastrous, as illustrated by Bloodstein, who tried for years to implement the treatment suggested by the theory, but failed completely (see Bloodstein, 1986). I do not think my concerns are an overstatement. A theory of the cause and development of stuttering would certainly be fine as a basis for treatment as you say, but with two provisos. First, the theory is verified with the scientific method. Second, the treatment based on the theory is evaluated with clinical trials of an acceptable standard. Surely it is unethical to provide health care with unproven treatments? The local doctor would not do it for asthma, and neither should the local speech pathologist do it for early stuttering.
Yes, the effectiveness of the Lidcombe program at the population level is not yet well researched. Also, it may not be the only way to treat early stuttering. There might well be other treatments, and I look forward to the publication of clinical trials of other treatments. If clinical trials show that there is a better and quicker treatment, I will be the first to use and endorse it.
However, I would not endorse multifactorial theory of stuttering as a source of treatment development. First, multifactorial theories of stuttering are completely and irretrievably wrong from empirical and logical perspectives. Second, no clinical trial shows the capacity of multifactorial treatments to control stuttering. There is a real risk that the errors of the diagnosogenic era will be repeated if we use multifactorial theories to treat children: that we will think that we know the cause of stuttering when we do not, and that we know what is an efficacious treatment for early stuttering when we do not.
I also do not agree with your claim that one type of theory, such as multifactorial theory, has been more successful than another such as single factor theory. In fact, no theory has been successful in explaining the cause of stuttering (see Packman & Attanasio, 2004).
Regarding future theories, you imply that history must repeat itself; I only say that it might. To prevent this, we both highlighted safeguards, such as that the theory must generate testable hypotheses. Meanwhile, it is true that the data for other treatments are presently lacking - much work remains to be done. To accomplish this, I would like to have a framework to support that therapy and the necessary data collection.
Single-factor or uni-dimensional explanations do not provide an appropriate starting point, for stuttering encompasses more than just speech disruptions, and the factors involved in the onset and development of the disorder involves more than just speech. Thus, I would challenge your claim that "multifactorial theories of stuttering are completely and irretrievably wrong." Regardless, it might help if we differentiate between multifactorial theories and resultant (or non-resultant) treatments. Obviously, theories cannot be used to treat stuttering, but they can provide the basis for a testable treatment.
As we move toward collecting needed data, we can start with a well-constructed theory that explains why we might manipulate certain variables. That would lead to studies, including clinical trials, that demonstrate the validity and efficacy of treatment. Lidcombe has accomplished this from an atheoretical perspective. I believe the same goal could be accomplished (though this has not yet been done) from a theory-driven perspective.
Yes, a new theory can lead to a new treatment checked by clinical trials. But not through a multifactorial theory.
There is nothing wrong with multifactoriality. Life is multifactorial, so is love, a toothbrush, and so is stuttering. However, arguing that because stuttering is multifactorial, therefore the cause of stuttering must also be multifactorial is an error of logic. This logical fallacy is called the representativeness reasoning: to believe that the nature of effects reflects the nature of causes (see Onslow, Attanasio, & Packman, 1998).
Additionally, multifactorial theories do not do what a theory should do: explain things (Packman & Attanasio, 2004). Why do word and syllable repetitions predominate at the onset of stuttering? Why can stuttering be intractable for a lifetime? Why do those who stutter have problems with tapping finger sequences? And why do those who stutter have the problem while playing wind instruments?
I found your list of relevant factors to be enlightening. In developing a theory to explain the phenomena of stuttering, it is appropriate to begin by listing those phenomena. Here are some questions that strike me: Why does stuttering start at a time of rapid expansion in linguistic, motoric, and temperamental aspects of children?s development? Why do people who stutter react in the ways they do? Why is stuttering so closely linked with language planning in children and adults? Why is stuttering not distributed randomly with respect to linguistic, situational, and experiential variables? Why do people who stutter show differences in motoric stability and linguistic processing, even when they are not speaking? What about differences in neural function and possibly even structure? And temperamental differences? Finally, why are multiple loci implicated in genetic modeling?
This is not an exhaustive list of the questions that would need to be answered by a comprehensive theory. Posing a single factor to explain all of this would seem unlikely. Posing multiple, interacting factors gives the opportunity to save more of the phenomena.
Moving back to treatment, I would like to see further discussion of what might be happening in children who do not recover through Lidcombe. Also, what might be changing in children who do recover, whether it be through Lidcombe or through other therapies? This would enhance our understanding of the disorder from both a theoretical and clinical perspective, and it would provide better justification for why we do what we do in therapy. Until we understand the reason for any change in fluency, I will remain dissatisfied with the knowledge base and will seek to identify some means of explaining the many and varied phenomena of this disorder.
Clinical trials and cohort studies give me no reason to think that there is a group of children who do not respond to the Lidcombe Program. Jones et al. (2000) reported 261 treated children of which 250 completed Stage 1. Thus, 250 children attained zero or near-zero stuttering. The remaining 11 did not complete for reasons common in speech pathology treatments, such as moving away, illness, severe family problems. These findings were replicated by Kingston et al. (2003). A similar picture has emerged with the Phase I, II, and III clinical trials that have been published.
Still, questions remain about its real-world effectiveness, for not everyone may administer the program as efficiently or as effectively as you and your colleagues do. We have discussed non-responders before. You may not see them in your studies, but I know of clinicians who see them in their daily practice. Clinicians have consulted with your team and other Lidcombe practitioners, and your team is quite responsive in trying to help clinicians in such cases. So, such cases seem to exist.
As for other treatments, in a preliminary study of the approach used at our Stuttering Center (Yaruss, Coleman, & Hammer, 2006), we found that 17 out of 17 children enrolled in the program (not just those treated to completion), achieved improved fluency and maintained it over a follow-up period of at least 2 to 3 years. Most required only the 6 sessions that the program is designed around, but a few children required more treatment. Why did the 4 children require more than 10 sessions? Ultimately, these children also recovered, but understanding their lack of immediate success might teach us about the disorder. So, we look at factors like language skills, motor skills, temperament, family history, and life experiences to help us better understand the treatment and the disorder itself. Have you conducted similar inquiries with Lidcombe? What theoretical framework did you use?
Again, there is no scientific evidence for the existence of a substantial cohort of non-responders. Also, my position is that the population effectiveness of the Lidcombe Program is currently unknown to science, because no effectiveness research has been conducted. But we have evidence for its efficacy. Epidemiological studies would be needed to address the capacity of the Lidcombe Program to impact at the clinical population coalface. We have put in place various ways to facilitate that effectiveness. For example, a Lidcombe Program Trainers Consortium has been established in seven countries (see the ASRC website http://www.fhs.usyd.edu.au/asrc). Each year, hundreds of clinicians around the world receive training from Consortium clinicians who meet published scientific benchmarks for the treatment.
If you know of someone who consistently does not get children to stop stuttering with the Lidcombe Program, there are at least two possible reasons. First, they may have not done the treatment according to the manual that can be downloaded from our website. Second, they may benefit from Consortium training in the treatment.
You accept that the LP has the best clinical trials efficacy research. So, what treatment do you select for a four-year-old stuttering child in need of treatment?
I treat as described in Yaruss et al. (in press), but I also work to validate that treatment and collect data to refine and improve the treatment. Though I do not use Lidcombe, my staff has received training and we have discussed the principles of the treatment in detail. Furthermore, I always encourage clinicians to participate in the Consortium training directly if they are considering using the Lidcombe program. I fear that many might not be using the treatment correctly, and without an understanding of why the treatment should work, it is impossible to know what the results from various modifications might be. This is yet another reason I keep returning to the value of theory, and why I asked the question that started this dialogue.
For my part, before I accept a treatment such as Lidcombe, I want to have a better idea about the nature of the changes that are taking place. Thus, in my clinic, we are actively engaged in examining not only the efficacy of the treatment we use, but also the mechanisms behind the observed changes. Much more work remains to be done, but our work, and that of others, is progressing.
Acknowledgements: Thanks to Mark Onslow and Scott Yaruss for the debate, to Tom Weidig, for editing on behalf of the BSA research committee, and to Mary Anne Strange, independent speech and language therapist, who suggested the idea of a debate.
Author Details: Associate professor Mark Onslow is the director of the Australian Stuttering Research Centre at The University of Sydney. His research interests include clinical outcomes of stuttering treatment, theories of proximal and distal causes of stuttering, and speech motor function in stuttering.
Associate professor Scott Yaruss is the clinical research consultant at the Children's Hospital of Pittsburgh and co-director of the Stuttering Center of Western Pennsylvania. His research focuses on identifying linguistic and motoric factors that influence the development of fluency and stuttering in children, and on improving the diagnosis and treatment of people who stutter.
From the Summer 2006 edition of Speaking Out, pages 17-19