Also see But Wait, There's More by Nan Bernstein Ratner, Ed.D.
By Ehud Yairi, Ph.D.
University of Illinois and Tel Aviv University
An article entitled Natural History of Stuttering to 4 years of age reporting a study conducted in Australia by Reilly, Onslow, Packman, et al. (2013), was recently published in the journal Pediatrics and has caused an unusual wave of letters, some strongly worded, from puzzled scholars, concerned clinicians, and confused parents, as well as calls from journalists. Whereas scientific issues are typically handled in scholarly outlets, the level of reaction appears to warrant a response in a venue that reaches a wide ranging audience. I thank the Stuttering Foun-dation  for offering these pages for comment.
Beginning with the positive, the general method employed - a longitudinal investigation of stuttering in a good-size sample, representing the general population of an area, starting in very early childhood, and employing multiple variables, is an excellent one to pursue. It allows for obtaining a reasonable estimate of the incidence of stuttering, documenting the onset close to its occurrence, as well as tracking the further development of the disorder. In this study, parents of approximately 1,600 children agreed to contact the investigators if they suspected that their child began stuttering. As a result, 181 children were identified and confirmed by the investigators as exhibiting stuttering onset.  Initial, first-year results were reported. With time, more data on recovery and persistency are expected.
Unfortunately, the authors make peculiar statements, overlook previous research, err in reporting findings from other studies, and reach conclusions that are beyond the scope of and/or not supported by their data. 
Reilly et al. state that “There is limited information about onset and recovery rates to inform practitioners….” Actually, there have been no fewer than 12 studies on onset of stuttering that included thousands of children and their parents. Most unsettling is Reilly et al.’s failure to mention the largest series of investigations, collectively known as The Iowa Studies, directed by the late Professor Wendell Johnson, the world renowned pioneer in this area. His name does not even appear in the article. Incidentally, he investigated a good number of the variables used by the Reilly team. Similarly, there have been numerous reports on recovery that have provided abundant information (see reviews by Wingate, 1976; Yairi & Ambrose, 2005; 2012). The unmentioned important references aside, one wonders what specific missing information was, and is being, sought by this team.
Reilly et al. state that their study “has clarified 3 important unknowns about the population epidemiology of early childhood stuttering.” Actually, none of these findings is an “unknown,” and stating that they have been “clarified” is an unbefitting and hasty proclamation.  In general, the authors make statements like this that border on the sensational rather than the factual or scientific.
First finding. The authors declare that their incidence of stuttering (11.5%) is higher than  figures published in the past, supposedly because they identified many early cases missed by other studies that did not assess children until they were 3 years of age or older. NOT SO. The literature clearly reveals that even higher incidences, such as 15% and 17%, were reported by Glasner and Rosenthal (1957) and Mansson (2005) respectively. The two references deserve credit. Also, other investigators included data for children under age 3 (e.g., Andrews & Harris, 1964; Craig, et al, 2002). Although 21st century studies have tended to yield higher figures than the commonly accepted 5% (see the Yairi & Ambrose, 2012 review), and I have defended higher figures, claiming that the incidence of stuttering has been “clarified” is premature. The missing credits aside, considerably more data on large samples of wider age range in various global locations are required for this to be accomplished.
Second Finding. The authors report that their children who stuttered displayed better language skills than nonstuttering counterparts. This too is not new. Already in 2001, Hage, in Germany, reported that preschool age children who stuttered demonstrated language skills at, or above, age expectations. In an even earlier American study, Watkins and Yairi (1999) concluded that “Persistent and recovered stutterers displayed expressive language abilities near or above developmental expectations…Children who entered the study at the youngest age level consistently demonstrated expressive language abilities well above normative expectations…(p. 1125). These missing citations of similar data by other researchers aside, the Australian finding certainly falls short of “clarifying” language as a factor in stuttering onset. To begin with, group means obscure those children whose scores are really high and those whose scores are low. The questions remain- who are these children and in what ways, if at all, do their high or low language skills affect their stuttering, at present, and over time? Are we dealing with subgroups?  Interestingly, Nicoline Ambrose and her team are producing data showing that, language-wise, those who persist trail those who recover. Another serious possibility (or likelihood) is that language skills are (a) an irrelevant parameter that (b) has been investigated with weak instruments. Indeed, McPherson and Smith (2013) have just shown that what might matter more is the level of language complexity. In an investigation of the motor speech production abilities of preschoolers who stutter, they found that abnormal patterns of motor activation were more likely when linguistic complexity increased. In short, it is imprudent of our Australian colleagues to suggest that the unknown role of language in early stuttering has been clarified.   
Third Finding. The authors reported that the brief, parent-report instrument they used to obtain a general indication of mental health, temperament, and psychosocial health showed no or little evidence of harm to the children in these domains. On this basis they offered the baffling conclusion that there is an absence of harm in any of the above areas and, the even more baffling interpretation, that “presence of stuttering improves those outcomes.” In other words, they say that during its first year, stuttering does a lot of good for the children! This is pretty amazing to me. Years prior to this study, other investigators, again not mentioned in the article, wrote that temperament differences “…were not reflected in the reports of children’s behavior as a group” (Yairi & Ambrose, 2005, p. 263).  Importantly, we certainly did not entertain the conclusion that such characteristics are absent in those who stutter.  Investigators and clinicians must be cognizant of the very significant limits of the instruments employed by Reilly et al. and their power to assess little children’s thoughts, feelings, reactions, etc., not to mention their insensitivity to the limitations of group means, especially in a heterogeneous population like those who stutter. Additionally, one must remember that the data represent parents’ opinions, not direct evaluation of children’s responses. Those who bothered to check Johnson et al. 1959  data, noted the opposite finding - that “...On 35 of the 36 items… the control group children were rated more favorably by their mothers or fathers, or both, than were the experimental group children by their parents (p. 57).” Also, subsequent researchers (e.g., Anderson et al., 2003) found indications of negative emotion and emotional reactivity.  Inasmuch as contradictory findings have been reported, it is too early to suggest that emotional/social aspects of young children who stutter have been clarified. Parents and clinicians should watch for overt manifestations or hints of such reactions.   
The Australian team also reported that stuttering tended to occur more often in homes where mothers had higher education than mothers of control children. Readers would have benefited had the authors compared their findings to those of Boyle, et al. (2011) who, in a very large survey sponsored by the USA Center for Disease Control, found just the opposite.
People have reacted particularly strongly to the authors’ recommendations that “Current best practice recommends waiting for 12 months before commencing treatment, unless the child is distressed, there is parental concern, or the child becomes reluctant to communicate. It may be that for many children, treatment could be deferred even longer.” This is not a logical conclusion. Inasmuch as the authors are tuned only to their own data showing only 6% natural recovery after one year, a logical recommendation would  have been for immediate clinical intervention for all children soon after onset, an observation wisely made by Donaher and Kelly (2013). Thus, the study does not present a scientific justification for a one-year waiting prescription. Oh, yes, perhaps they rely, without saying so, on Yairi and Ambrose (2005) who opined that waiting 6-12 months or so after onset is an option. But, we investigated and found high recovery over several years and also reported the significant progress of children during the first year to justify our conclusion. Furthermore, we also advocated a selective approach, offering a variety of data-based predicting factors for risk of persistent stuttering that call for early intervention. For example, a child who is a boy, has a family history of persistent stuttering, and shows no improvement over several months, presents with a high risk for persistence. In this case then, there is appropriate justification for immediate initiation of therapy. It is puzzling and problematic that Reilly et al. seem to overlook their own data as well as other relevant developments in the field. Their progress report seems to be premature. 
Finally, there is indeed one very unique outcome reported by Reilly, Onslow, Packman, et al. They found that family history of stuttering is NOT a predicting factor for stuttering onset (8.3% of stuttering children had familial history vs. 5.5% in controls). In this respect their results contradict some 30+ studies that have found strong family connections to stuttering onset. This suggests that either their sample was rather atypical or, more likely, the method of securing this information was insufficient. Approximating my 75th, and with a bit of a smile, I entertain a third possibility - that down under also genetics swirls in the opposite direction.
Anderson, J., Pellowski, M, Conture, E., & Kelly, E. (2003). Temperamental characteristics of young children who stutter. JSLHR, 46, 1221-1233.
Andrews, G., & Harris, M. (1964). The syndrome of stuttering: Clinics in Developmental Medicine, No. 17. London: Heinemann Medical Books.
Boyle, C., Boulet, S., Schieve, et al. (2011). Trends in the Prevalence of Developmental Disabilities in US Children, 1997-2008. Pediatrics, 34,385-95.
Craig, A., Hancock, K., Tran, Y., et al. (2002). Epidemiology of stuttering in the community across the entire life span. JSLHR, 45, 1097-1105.
Donaher, J., & Kelly, E. (2013). Too much made of too little too soon. Pediatrics,
Glasner, P. J., & Rosenthal, D. (1957). Parental diagnosis of stuttering in young children. J S HD, 22, 288-295.
Johnson, W., & Associates. (1959). The onset of stuttering: Research findings and implications. Minneapolis, MN: University of Minnesota.
Månsson, H. (2005).  Stammens kompleksitet og diversitet. Dansk Audiologopædi, 41, 13-33.
Reilly, S., Onslow, M., Packman  et al. (2013). Natural History of Stuttering to 4 Years of Age: A Prospective Community-Based Study. Pediatrics, 36, 460-467.
Watkins, R. V., Yairi, E., & Ambrose, N. G. (1999). Early childhood stuttering III: Initial status of expressive language abilities. JSLHR, 42, 1125-1135.
Wingate, M. (1976). Stuttering: Therapy and treatment. New York, NY: Irvington.
Yairi, E., & Ambrose, N. (2005). Early Childhood Stuttering. Austin: Pro-Ed. 
Yairi, E. & Ambrose, N. (2012).  Epidemiology of stuttering: 21st century advancements.  JFD,   
From the 2014 Winter Newsletter