JCPSLP Vol 18 no 2 July 2016

characteristics. The most consistent predictor of relapse or regression is pre-treatment stuttering severity, usually measured as percentage of syllables stuttered (%SS). This has been found in earlier studies as well as more recent ones of predictive factors of treatment outcomes in AWS (Block, Onslow, Packman, & Dacakis, 2006; Craig, 1998; Guitar, 1976; Huinck et al., 2006; Ladouceur, Caron, & Caron, 1989). Of the more recent studies, Huinck et al. (2006) investigated subtypes of AWS ( n = 25) based on pre- treatment scores of stuttering severity (mild or severe) and severity of negative emotions and negative cognitive thoughts (mild or severe). Only stuttering severity was found to be a predictor of treatment outcomes; people with severe stuttering demonstrated the largest gains in therapy but they also experienced more relapse as measured by %SS (Huinck et al., 2006). Block and colleagues (2006) conducted a prospective investigation of predictive factors for treatment outcomes with short- and long-term (up to 5 years post-treatment) follow-up periods. Consistent with previous findings, pre- treatment stuttering rate (%SS) was found to be a predictor of short-term treatment outcome ( n = 78). That is, AWS with mild stuttering achieved better treatment outcomes, as measured by %SS, at the 1-year follow-up and were less likely to relapse than adults with a severe stutter. In turn, the only predictor of long-term treatment outcome was the short-term stuttering rate at 3 months post-treatment. Additional factors Caution is required when interpreting the research literature, given that studies of factors to predict relapse other than stuttering severity have yielded inconsistent results, are based on single studies, and/or are based on weak evidence. As such, their ability to predict treatment success is currently questionable and the findings need replication. These factors requiring further investigation include those related to communication attitudes, locus of control (internal or external), social anxiety, personality profiles (Iverach et al., 2010), mental health disorders (Iverach et al., 2009), and resilience, which are the processes and mechanisms by which an individual deals with adversity in life (Craig, Blumgart, & Tran, 2011). Craig et al. reported that AWS who were more resilient had lower levels of health risk, were able to manage their stress levels better, had fewer physical as well as social limitations, and had more vitality and social support. Factors found not to predict treatment outcomes Factors that have been examined but not found to predict treatment outcome include the following: age, gender social status, neuroticism, extroversion, avoidance, reaction (e.g., negative reactions to stuttering), self-help activities, formal practice (booster/maintenance sessions), real-life assignment (practice of techniques in a functional setting), post-treatment speech naturalness, number of first-degree relatives who stutter, and whether or not previous treatment had been received (Block et al., 2006; Craig, 1998). Children who stutter Early stuttering It is a surprise and a concern that very little data are available to provide information on the prediction of treatment outcomes for early stuttering. In 2000, Jones et al. stated that “almost nothing is known … about factors that predict the responsiveness of early stuttering to treatment” (Jones, Onslow, Harrison, & Packman, 2000, p.

1441). Jones et al. investigated factors including age, time since stuttering onset, gender, and stuttering severity in 261 children that could predict treatment duration for the first stage of the LP. They found that stuttering severity at the first treatment session was the only factor that related to the how long it took for children to enter the second (maintenance) stage of the program. Children with more severe stuttering required more time. There have since been more studies of factors predicting outcomes on the LP for early stuttering, with several studies consistently reporting that treatment takes longer for children who stutter more frequently or more severely pre-treatment (Jones et al., 2000; Kingston, Huber, Onslow, Jones, & Packman, 2003; Koushik, Hewat, Shenker, Jones, & Onslow, 2011). A follow-up investigation of the same children from Jones et al. (2000) found that majority of children treated with LP maintained their fluency gains (below 1% syllables stuttered) 5 years post-treatment (Jones et al., 2008). A minority of the children relapsed, but the authors stated that it was unknown if there were any predictors of long-term relapse after treatment with the LP as their analyses did not yield useful results regarding this. More recently, a replication and re-analysis of the data from children treated by the LP (independent of the LP developers) was conducted by Guitar and colleagues (2015). The authors combined the data from two sets of children who had been treated with the LP ( n = 29) to investigate predictors of long-term outcomes of treatment with the LP. They tested pre-treatment stuttering %SS and gender as factors, and found that both were significant predictors. They found that females had better long- term outcomes as measured by %SS and that this was independent of their pre-treatment %SS. For males, long- term outcome in %SS was positively correlated with their pre-treatment %SS. Older children and adolescents For older children and adolescents, pre-treatment stuttering severity has been found to be a predictive factor for treatment outcome (Cook, Howell, & Donlan, 2013; Hancock & Craig, 1998). Hancock and Craig (1998) (n = 77) found that trait anxiety post-treatment could also predict stuttering frequency 1-year post-treatment. Psychosocial measures and a language measure of lexical diversity did not predict stuttering severity following treatment (Cook et al., 2013). Interpreting predictive factors The most reliable factor in predicting treatment outcome is pre-treatment stuttering severity in AWS. To date, this seems to also be the case for children who stutter, although data is limited and has been reported mainly for the LP. Other client factors could be useful for predicting treatment outcomes, and more studies are required with improved research design and replication of results. Replication of results is also necessary across different sites independent of the original location and the program developers, for example, for programs like the LP (Guitar et al., 2015; Jones et al., 2000; Kingston et al., 2003; Koushik et al., 2011). Conclusions and future directions At present, the most consistent prognostic factors associated with stuttering onset – that is, factors to predict who is likely to start stuttering – are a positive familial history of stuttering, age, and gender (being male). The prognostic factors related to natural recovery – that is, who is likely to

97

JCPSLP Volume 18, Number 2 2016

www.speechpathologyaustralia.org.au