An evaluation of the effectiveness of the Lidcombe program of early stuttering intervention

This thesis presents a randomised controlled trial of the Lidcombe Program of Early Stuttering Intervention. The Lidcombe Program was developed for the treatment of stuttering in preschool-age children. The effectiveness of the Lidcombe Program was compared to a control group in a parallel group randomised controlled trial with blinded outcome assessment. A number of supplementary studies were conducted in support of the trial; two literature reviews, two retrospective file audits and a statistical simulation study. A review of randomised studies of treatments for stuttering showed that there have been 27 such studies published in English language journals. Of these only one was devoted to a treatment for early stuttering and that was the Lidcombe Program. The randomised study showed that 3 months of this treatment was associated with a lower level of stuttering compared to a control group who received no treatment. However, with a sample size of 23, this study lacked power and the children did not receive a full course of treatment. Despite these limitations, this study provided evidence that a medium to large effect size could be anticipated in an adequately powered and properly conducted randomised controlled trial. The second review was of sample size and power in stuttering research studies that had been published in two speech pathology journals; the Journal of Speech, Language and Hearing Research (Vol 39, No. 1 to Vol 40, No, 4) and the Journal of Fluency Disorders (Vol 21, No. 1 to Vol 22, No, 3). Results suggested that the majority (73%) of the 26 studies reviewed were insufficiently powered to detect even large effects. However it was acknowledged that it is very difficult to recruit even moderate sample sizes of people who stutter. It was concluded that one way to help improve this situation is collaboration of multiple research centres or, in the case of a randomised controlled trial, inclusion of multiple recruitment sites in one study. This strategy was adopted in the randomised controlled trial reported in this thesis. Two retrospective file audit studies of children treated with the Lidcombe Program were conducted in Australia and Britain. One purpose of these file audits was to obtain information relevant to the design and conduct of the randomised controlled trial. Data from the case reports on more than 300 children from the two sites were included in a meta-analysis. Results showed that a median of 11 weekly clinic sessions were required for children to attain the criteria for low levels of stuttering for completion of Stage 1 of the Lidcombe Program. Approximately 90% of children had achieved those criteria within 6 months of beginning treatment and almost all children had achieved them within 1 year. In addition two variables were found to be associated with longer treatment duration: more severe pre-treatment stuttering and shorter times from onset of stuttering to the start of treatment. The latter was apparent in the meta-analysis but not for the individual cohorts. As a result of these findings, pre-treatment stuttering severity was stratified along with other relevant variables in the randomised controlled trial and follow up for participants was a minimum of 9 months. A simulation study was conducted prior to analysis of data from the primary outcome measure of the randomised controlled trial: percentage of syllables stuttered (%SS). The distribution of %SS scores is positively skewed. Nonetheless, simulation showed t-test to be an appropriate analysis for this primary outcome measure. There were two treatment sites for the randomised controlled trial: the University of Canterbury (Christchurch, New Zealand) and the Stuttering Treatment and Research Trust (Auckland, New Zealand). A total of 54 preschool-age children were recruited: 29 to the Lidcombe Program and 25 to the control group. Half the proposed sample size was achieved due to slower than anticipated recruitment. This occurred because, as the trial progressed, treatment with the Lidcombe Program became common knowledge among parents in New Zealand and they became increasingly reluctant to agree to have their child randomised to the trial. Analysis with t-test showed a highly statistically significant difference (p = 0.003) at 9-months post-randomisation. The mean percentage of syllables stuttered (%SS) at 9-months post-randomisation was 1.5 (SD = 1.4) for the Lidcombe Program group compared to 3.9 (SD = 3.5) for the control group, resulting in a treatment effect of 2.3 %SS (95% confidence interval: 0.8-3.9). This treatment effect was more than double the minimum clinically worthwhile difference specified in the trial protocol. These results show that the Lidcombe Program is significantly more effective than natural recovery for reducing stuttering levels in preschool children. The Lidcombe Program is the first early stuttering treatment to be shown to be more effective than natural recovery in a randomised controlled trial