Non-pharmacological treatments for stuttering in children and adults: a systematic review and evaluation of clinical effectiveness, and exploration of barriers to successful outcomes

Search strategies

A systematic and comprehensive literature search of key health, medical and linguistic databases was undertaken in August 2013 to February 2014. The searching process aimed to identify studies that reported the clinical effectiveness of interventions for people who stutter and also studies that reported the views and perceptions of people who stutter and staff regarding interventions. Searching was carried out for both reviews in parallel, with allocation to either effectiveness or qualitative reviews at the point of identification and selection of studies for potential inclusion. The search process was recorded in detail with lists of databases searched, date search run, limits applied, number of hits and duplication as per PRISMA guidelines. 7 The search strategy is presented in Appendix 1.

The search involved combining terms for the population (stuttering) with terms for the interventions of interest, that is, non-pharmacological interventions. This highly sensitive search strategy (i.e. not using terms for comparators, outcomes or study design) was possible because scoping searches retrieved relatively small and manageable numbers of citations. The aim of the strategy was to identify all studies on non-pharmacological interventions for stuttering.

The search strategy was developed by the information specialist on the team (Anna Cantrell) who undertook electronic searching using iterative methods to create a database of citations using Reference Manager version 12 (Thomson ResearchSoft, San Francisco, CA, USA). The search followed a process whereby search terms were developed initially from scrutinising relevant review articles, followed by scrutinising retrieved papers to inform further searching.

The first main project search was run on MEDLINE (via Ovid) and PsycINFO (via Ovid) in August 2013. Following minor amendments to the search terms, a further iteration of the search was then conducted on a larger range of databases in October to November 2013. Topic experts and clinicians in the field were consulted for additional search terms and for suggestions of additional relevant studies or interventions at regular advisory group meetings and at a clinician workshop session.

In addition to standard electronic database searching, later in the project (February 2014) citation searching was undertaken for all included qualitative citations and searches were conducted for additional papers by the first authors of all included qualitative studies. In order to ensure that the most up-to-date literature was not missed, we also conducted hand screening of journals in April 2014 to identify any work published since the main searches had been carried out. The journals that we searched by hand were International Journal of Language & Communication Disorders; Journal of Speech and Hearing Research; Journal of Communication Disorders; Asia Pacific Journal of Speech Language and Hearing; Clinical Linguistics and Phonetics; Journal of Fluency Disorders; and International Journal of Speech-Language Pathology.

Sources searched

The following electronic databases were searched for published and unpublished research evidence from 1990 onwards.

First search iteration

• MEDLINE (via OvidSP).

• PsycINFO (via OvidSP).

Second search iteration

• EMBASE (via OvidSP).

• Cumulative Index to Nursing and Allied Health Literature (EBSCOhost).

• The Cochrane Library (Wiley) including the Cochrane Systematic Reviews Database, Cochrane Controlled Trials Register, Database of Abstracts of Reviews of Effects, Health Technology Assessment and NHS Economic Evaluation Database databases.

• Applied Social Sciences Index and Abstracts (ProQuest).

• Linguistics and Language Behavior Abstracts LLBA (ProQuest).

• Science Citation Index (Web of Science).

• Social Science Citation Index (Web of Science).

• Conference Proceedings Citation Index – Science (Web of Science).

• Sociological Abstracts (ProQuest).

• Evidence for Policy and Practice Information and Co-ordinating Centre databases.

All citations were imported into Reference Manager and duplicates deleted prior to scrutiny by members of the team.

Search restrictions

Searches were limited by date (1990 to present) as the advent of new programmes may have led to changed practice and the review was aiming to synthesise the most up-to-date evidence. This date criterion was set as it marked a major change in interventions for stuttering associated with publication of the first papers reporting the Lidcombe Approach, with the field from this date forward addressing the need for more public evidence for effectiveness. The review thus encompassed nearly 25 years of research.

The searches did not set an English-language restriction. Although we intended that the review would be predominantly limited to work published in English to ensure that papers were relevant to the UK context, we aimed to search for and include any additional key international papers.

Population

• The population eligible for inclusion was a person who stutters of any age. This included those with overt stuttering behaviours such as repetition of syllables or blocking, those with covert behaviours such as word avoidance and also those diagnosed with any other disorder of developmental fluency such as cluttering.

• The review excluded people with a fluency disorder which had been acquired rather than developmental, such as non-fluency associated with an identified neurological impairment (such as head injury, stroke or Parkinson’s disease).

• We included studies whose participants were described as being clutterers. Although cluttering is considered a distinct disorder from stuttering, it is recognised in the field that it may be challenging to differentially diagnose, and can also co-occur with stuttering. Therefore, we took the decision to search for and include any literature meeting our criteria, which examined interventions for this population. However, this work would be highlighted in the results as a separate population group.

• The review excluded papers reporting interventions for children who have been defined as having normal non-fluency by the authors of the source study.

• The qualitative review considered studies reporting the views and perceptions of interventions for stuttering. The population was people who stutter, their relatives, friends or significant others, together with the views of staff delivering interventions.

Interventions

• The review included any intervention that had the stated aim of being of benefit to people who stutter. This could be by either reducing the frequency of occurrence of behaviours (overt and/or covert), or by aiming to address communication and/or social restrictions.

• Non-pharmacological interventions were included.

• Interventions delivered in any setting by any agent were included. This encompassed treatments provided as part of state-funded health service provision, those offered by private providers and interventions delivered by charitable or voluntary organisations.

• The review excluded interventions that are pharmacological.

• The review excluded interventions that do not have the stated aim of improving fluency outcomes, for example general relaxation or massage sessions, or the provision of information about stuttering.

Comparators

• Studies with any comparator including an alternative intervention, no intervention or usual practice were eligible for inclusion. This included studies that compared pharmacological to non-pharmacological intervention.

• Studies comparing pharmacological intervention to no intervention were excluded.

Outcomes

• Any outcome relating to a positive effect on the communication or emotional well-being of people who stutter was included.

• Relevant outcome measures included test scores on a standardised assessment such as frequency of non-fluent words; patient self-report of covert stuttering; patient experience; report of frequency of stuttering from a significant other such as a teacher or employer; and patient or staff views and perceptions of obstacles to intervention effectiveness.

• Outcomes related to reading aloud only, rather than any measure of communicative interaction were excluded.

Study design

• The review included designs which may be termed randomised controlled trials (RCTs), randomised cross-over trials, cluster randomised trials, quasi-experimental studies, cohort studies, before-and-after/longitudinal studies, case–control studies and non-survey cross-sectional studies.

• Case reports (a single participant), case series (defined as reporting data from two or three participants) and survey (questionnaire) study designs were excluded.

• The qualitative review examined studies that reported the views of people who stutter or staff perceptions. Any qualitative method was eligible for inclusion (such as interviews and focus groups). Non-qualitative data collection methods such as questionnaire/survey designs were excluded.

Other inclusion/exclusion criteria

• The review included studies from any Organisation for Economic Co-operation and Development (OECD) country, thus studies from non-OECD countries were excluded.

• Studies published in English and key studies published in other languages were included. Studies published in languages other than English without an English abstract were excluded. Studies published in languages other than English which had English abstracts were considered; however, only those considered to be key studies that may add significantly to the review (based on the information in the abstract) were eligible for translation and inclusion.

• Grey literature (unpublished evaluations) from the UK was eligible for inclusion.

Effectiveness studies

Data were synthesised in a form appropriate to the data type. It was proposed that meta-analysis calculating summary statistics would be used if heterogeneity permitted, with use of graphs, frequency distributions and forest plots. It was anticipated that subgroups including age of participants, learning disability, intervention content and delivery agent would be examined if numbers permitted. However, the heterogeneity of the included work precluded summarising the studies via meta-analysis.

Clinical effectiveness review findings were reported using narrative synthesis methods. We tabulated characteristics of the included studies and examined outcomes by typologies, outcome measurement, intervention dosage and length of follow-up. Relationships between studies and outcomes within these typologies were scrutinised.

Qualitative studies

Qualitative data were synthesised using thematic synthesis methods10 in order to develop an overview of recurring perceptions of potential obstacles to successful outcomes within the data. This method comprises familiarisation with each paper and coding of the finding sections (which constitute the ‘data’ for the synthesis), according to key concepts within the findings. Although some data may directly address the research question, sometimes information such as barriers and facilitators to implementation has to be inferred from the findings, as the original study may not have been designed to have the same focus as the review question. 10

Study design

Table 3 details the included effectiveness papers categorised by study design. We have provided a definition of each category in order to ensure clarity. The reporting of study design used by authors encompassed a variety of terminology, with terms in some instances not accurately representing the true design. Fourteen papers16–29 reported studies with a comparator, of these four17,18,24,25 randomly allocated participants to each arm of the study, six16,19–23 allocated participants using quasi-randomisation methods (such as consecutive randomising) and one29 was a controlled before-and-after study with no allocation. Of these 14 papers, three reported data from the same study16,20,21 with the greatest proportion of included empirical work using a before-and-after design (pre- to post measure).

Although 26 studies carried out outcome assessment immediately following the intervention,4,19,22,26,34,35,42,43,48,52,54,60,65,67–69,83,93,111,118–120,123–126 there were 51 papers reporting follow-up periods of 12 months or more16,20,21,25,27–29,37,38,40,41,45–47,49,51,56,57,59,61,63,66,69,71–74,77,79–82,84,90,91,96,97,100,103,104,106–110,112,114–117,122 (Table 4).

Country of origin

A categorisation of included studies by country of origin is presented in Table 5. The greatest proportion of work was reported by authors based in Australia (39 papers), followed by the USA (26 papers). Eight papers were from the UK.

Intervention dosage

We endeavoured to identify from author report how many hours of intervention were provided in the included studies (Table 6). Papers varied considerably with regard to the level of detail provided and, therefore, the table below may not be completely accurate in representing intervention dosage, but is based on information we could glean. It can be seen that a sizeable proportion of the papers varied the number of hours of intervention according to individual need. This makes comparing effectiveness by dosage unfeasible. It can also be seen from the table that the contact time ranged from fewer than 10 hours to more than 75 hours, again making the drawing of comparisons between different interventions on the basis of dosage problematic. The interventions that had shorter contact times tended to be those which were based on the use of technology (such as DAF systems). The interventions with longer contact time (perhaps unsurprisingly) tended to be those with multiple elements.

Intervention provider

In terms of the person delivering the intervention, 51 studies reported that clinicians provided the therapy. In all except three cases these clinicians were speech and language pathologists/therapists (two interventions were delivered by clinical psychologists and one jointly by a therapist and psychologist). Fifty papers16–36,51–54,56–60,65–72,79,82–90,100–102 were unclear with regards to who delivered the sessions; it was presumed that in most cases this was the author/s. Eleven studies reported that student clinicians had been used to provide therapy, with supervision by qualified staff.

Cluttering

As outlined earlier, we took the decision to search for and include any work that examined interventions for people who clutter – a related speech fluency difficulty. We found only one paper which met our inclusion criteria and identified some of the participants as people who clutter. 72

Feedback and technology interventions

Twenty-two papers were included that described the effectiveness of a range of a technologies aiming to reduce the frequency or severity of stuttering in speech (Table 8). The earliest of these papers was published in 1996, and the most recent in 2014, with 13 of the papers from North America (see Table 8). The greatest proportion of the technologies described were devices that alter the way that a person who stutters hears their own speech [altered auditory feedback (AAF)] by changing the frequency [frequency altered feedback (FAF)] and/or by introducing a delay before the speech is heard (DAF). All but one of the included studies52 either compared stuttering level while using a device with stuttering level with no use of the device, or compared fluency level using different device settings. The other paper52 compared use of a device in people who stutter with use by non-stuttering speakers. All but one of the papers92 in this group was rated as being at higher risk of bias. The papers described the use of AAF under a variety of conditions including reading, monologue and conversation (either in person or via the telephone).

This type of intervention alters the auditory feedback process in people who stutter with the aim of reducing the proportion of stuttered speech. Although the precise area of change and way that these interventions act to reduce stuttering is debated, it has been proposed that they may activate a ‘mirror neural system’ to link perception with production or, alternatively, that they have an impact on timing processes that control speaking rate. In the following synthesis we have detailed only the findings relating to conversational interaction (or monologue if no conversational measure was available). Many of the papers contained further detailed data regarding outcomes in terms of reading aloud. 4,17,35,36,48,52,56,60,87,92,102,103,105,118–120,124–126

Use of the SpeechEasy device (Janus Development Group, Inc., NC, USA) was reported in six papers. 4,52,56,87,92,120 These studies explored the use of the technology in laboratory, clinical and naturalistic contexts and examined follow-up for periods up to 59 months. Sample sizes ranged from seven to 31 individuals with no studies using a control group design. Five out of the six papers were assessed as being at higher risk of bias, with only one92 judged to have a lower risk of bias.

All studies reported some degree of effectiveness for this intervention. Armson et al. 4 found stuttering was significantly reduced having the device in place versus no device (p = 0.01) with a small effect size (ES) of 0.108. However, there was considerable individual variation in responses, with the suggestion that those having lower initial stuttering had better outcomes. A second paper by Armson and Kiefte120 also reported significant decreases in stuttering rate with SpeechEasy compared with stuttering rate without SpeechEasy for all but two of 31 participants (p < 0.001, ES 0.724). The mean stuttering frequency pre-device was 16.4 and with device the mean was 2.3, an average reduction during monologue of 60.7%. Participant self-rating of stuttering severity also improved during the device condition (from 5.95 to 3.29; p = 0.028, ES 0.658). The paper examined whether or not stuttering reduction was at the expense of reduction in speech naturalness or rate and concluded that participants had a slower than normal rate both with and without the device. Naturalness ratings increased to just below normal levels with the device. The Foundas et al. 52 paper echoes these findings, with a significant reduction in stuttering frequency with the SpeechEasy device in place and activated versus in place but not producing DAF or FAF (p = 0.014, a 36.7% reduction). The paper examined the effect of different device settings and concluded that the setting preferred by the participants was more effective than the default setting. In contrast to the findings above, individuals with more severe stuttering at baseline had a greater benefit.

Three papers examined longer-term outcomes of SpeechEasy intervention. 56,87,92 One56 followed up device users following initial fitting. Eight of the 11 participants were still using the device at a mean of 37 months’ follow-up. The study found that level of dysfluency (for the seven participants that data were available for) was not significantly different at long-term follow-up than it had been at first fitting (p = 943). However, there was significant variation with three having increased fluency, one was unchanged and three had worsened fluency since initial fitting. Analysis of data for all 11 people who stutter (those who continued to use the device and those who did not) found that all had significantly improved levels of fluency from before they were fitted with SpeechEasy to the current time point (p = 0.017). The authors suggested that this indicates carry over effect from the device even when use discontinues. However, an alternative interpretation may question the long-term value of using the device in that continued users did not differ from non-users. In support of this, the study reports that at time of follow-up there was no difference in fluency whether the device was worn or not worn (p = 0.92).

The second paper reporting longer-term follow-up data92 similarly casts some doubt on the long-term clinical effectiveness of SpeechEasy and this paper was judged to be at lower risk of bias. This study examining beyond-clinic data found a positive effect on the percentage of syllables that are stuttered in the shorter term following fitting (p = 0.02); however, no significant effect on the percentage of syllables that are stuttered at 4-month follow-up (p = 0.090). Self-report scores on the Stuttering Severity Index (SSI) and Overall Assessment of the Speaker’s Experience of Stuttering Questionnaire (OASES) showed no difference pre- to post intervention; however, the Perceptions of Stuttering Inventory (PSI) scores had significantly improved (p < 0.05). Only 4 out of the 11 participants had purchased the device, eight reported they disliked the irritating background noise and five that they disliked being unable to hear self/others. Six reported that using the device had increased their confidence in speaking and six reported that they had an overall increase in fluency using it.

The O’Donnell et al. 87 paper includes beyond-clinic measures using data obtained via the telephone. This study followed participants at regular intervals for 16 weeks after fitting and included speech data and participant self-report. Use of the device varied from 2 hours per day to 15 hours per day. Stuttering reduced for all participants at the baseline evaluation point (by 75.5–97.9%); however, there was considerable variation in outcome between participants at the final follow-up. Four stuttered less with the device than those without it and three stuttered less without the device than those with it. Five of the seven stuttered more at follow-up than they had at baseline with the device in use (although all had reduced levels of stuttering when not using the device than they had previously). Analysis of the beyond-clinic telephone recordings indicated positive outcomes for five participants, with mean reduction in stuttering ranging from 20% to 94.4% when conversing with the experimenter while having the device in place, compared with not using it. On self-report measures, six participants described reduced struggle or avoidance behaviour with five participants identifying substantial benefit.

Six papers reported the use of other feedback devices combining DAF and FAF. All were considered to be at higher risk of bias. 35,103,104,118,124,125 Antipova et al. 118 used The Pocket Speech Lab (Casa Futura Technologies, CO, USA) with eight participants and found all reduced the percentage of words stuttered using the device by an average of 3–4%. The paper details individual response under eight different AAF conditions with a significant difference between these and the no-device condition (p = 0.049) in terms of the percentage of syllables that are stuttered. The authors report a trend for those with more severe stuttering to have a greater reduction; however, they highlighted the significant individual variability in response. Unger et al. 125 found a significant reduction in SSI severity rating (p = 0.000) for 30 participants using the VA 601i Fluency Enhancer (VoiceAmp Ltd , Middlesex, UK) or the SmallTalk devices (Casa Futura Technologies, CO, USA). Individual variability in outcome was also emphasised in this study. The Digital Speech Aid (Digital Recordings – Advanced R & D, Nova Scotia, NS, Canada) was evaluated in a study with a larger sample of 335 individuals. 124 Statistically significant improvement in the number of dysfluent syllables was observed using the device than non-use (p < 0.005). In dialogue, the odds ratio (OR) of exhibiting dysfluency without the device was 0.58 and with the device in use was 0.18. Although moderate or considerable improvement was found for 84.5% of participants, deterioration or lack of improvement was found for 15.5%.

Use of the Edinburgh Masker (Casa Futura Technologies, Boulder, CO, USA) in both clinic and home settings was evaluated by Block et al. 35 Results for the 18 participants showed a decrease in the percentage of syllables that are stuttered for all across all conditions (conversation with experimenter: 2.1% reduction of syllables that are stuttered, conversation familiar person: 2.6% reduction of syllables that are stuttered, telephone: 2.8% reduction of syllables that are stuttered). The authors reported that an analysis of variance (ANOVA) was performed which indicated a significant reduction in stuttering; however, the details of this are not provided. Some individual differences in response are described (eight participants increased stuttering on at least one task) and although speaking rate was found to be unaffected, speech naturalness appeared to be reduced using the device (p < 0.01).

Companion papers103,104 report 4-month and 12-month follow-up data from intervention using a self-contained in-the-ear prosthetic fluency device providing both FAF and DAF. The earlier paper103 describes three experiments using the equipment. The proportion of stuttered syllables was significantly reduced for the seven participants in experiment 1 when they used the device during monologue (p = 0.011, a 67% reduction of syllables that are stuttered). Similarly, for eight participants in experiment 2 there was a significant reduction in proportion of stuttered syllables (p = 0.0028). The third experiment focused on evaluating speech naturalness and found that speech while using the device was rated as more natural sounding than without (p < 0.0001), although scores were below that for normal speakers. The follow-up paper similarly outlines three experiments. The first found that initial reductions in stuttered syllables reported at initial fitting with the device in place compared with no device were repeated at 12 months (p < 0.0001), with a 75% reduction in percentage of syllables that are stuttered using the device during monologue. Experiment 2 details significantly improved PSI scores at 12 months compared with scores prior to receiving the device. Participants were asked to self-report current levels and recall previous, but this means the reliability of these data must be questioned. Experiment 3 examines speech naturalness and found an increased naturalness rating at 12 months compared with 4 months and that speech while using the device was rated as more natural than without (although as with the earlier paper was less natural than normal speakers).

Three papers focused on the use of AAF devices to reduce stuttering during use of the telephone. 44,60,126 The most recent paper60 examined the effectiveness of different combinations of DAF and FAF during scripted telephone conversations. Although this study could be perceived to be using a reading aloud only outcome and, therefore, falls within the exclusion criteria, the script was considered to be similar to notes that a person who stutters may make in everyday life when making a telephone call and so the study offered more functional outcomes. Stuttering frequencies in both AAF conditions for all nine participants were significantly lower than the non-altered feedback condition (p < 0.0001, an average of a 65% reduction). These findings are similar to an earlier paper126 which reported a reduction in stuttering frequency of 55–60% using AAF during scripted telephone conversations (p = 0.004) with a positive effect for all nine participants. Bray and James44 support the clinical effectiveness of using an AAF device when making telephone calls. The Telephone Assistive Device (VoiceAmp Ltd, Middlesex, UK) evaluated in this study reduced stuttering frequency for four out of five participants (group mean 8.28% pre-device and mean 4.82 using device). The authors suggested some improvement in self-reported feelings and attitude following use of the device, but there are limited data to support this.

One paper reported the use of FAF only,119 and another the use of DAF only. 105 Amson and Stuart119 found that although some improvement to reading using FAF was observed, there was no significant effect on the number of stuttering events during monologue, with 10 out of the 12 participants showing no benefit. Use of DAF over 3 months105 was found to significantly reduce the percentage of stuttered words (when using the device compared with not using it) for non-functional speech tasks and picture description (p = 0.050); however, not significantly for conversation (p = 0.066). Levels of stuttering without the device in place were significantly reduced from baseline levels for all but conversation (p = 0.0666). Overall levels of stuttering when using the device from baseline to 3-month follow-up had not significantly changed. Self-report perception of fluency (using median scores on the summary table provided) was that fluency using DAF was better than fluency without DAF for four out of nine participants (unchanged for four, worse for one).

Other types of technology evaluated in the literature were bone conduction stimulation and electromyography (EMG). Stidham et al. 101 reported the use of bone conduction stimulation with DAF which participants used for at least 4 hours a day for 4 weeks. Although baseline to immediate post provision of the device indicated a significant reduction in stuttering (p < 0.001), the effect had faded at 2- and 6-week follow-up. Of the nine participants, slightly more than half reported that their speech had improved using the device (56%) and 66% rated it as helpful to some degree. However, the headband element of the device was described as being uncomfortable and obtrusive.

Two papers examined the use of EMG feedback. One of these16 compared EMG with two other interventions and will be outlined in detail later (see Papers comparing interventions). In summary, this study found that for 6 out of the 10 children taking part, EMG reduced stuttering to less than 1% of syllables that are stuttered immediately post intervention, with four children remaining at this level at 1-year follow-up. The other paper36 used EMG with 12 children and adolescents daily over a 5-day period. There was a reduction of mean 36.7% in stuttering after treatment (pre-mean 4.9% of syllables are stuttered to post mean 4.4% of syllables are stuttered), however it was noted that rate of speech post intervention was only around half that of a non-stuttering population. One participant had a worse percentage of syllables that are stuttered following intervention.

The final papers included in this categorisation of feedback and technology interventions were three papers outlining the use video self-modelling (VSM) (participant viewing of videos of themselves which had been edited to remove stuttering). The self-modelling intervention tested by Bray and Kehle44 was carried out on seven occasions over 6 weeks. Results are reported descriptively by the four individual participants, with mean number of stuttered words ranging from 5.9 to 9.1 at baseline and 0.3 to 3.2 at the 8-week follow-up. A more recent paper48 evaluated the viewing of edited videos daily over a 1-month period. This study investigated the potential use of this intervention with people who stutter who had received previous interventions but had relapsed. Results indicated a significant reduction of 5.4% of syllables that are stuttered (p < 0.0001) post intervention, an ES of 1.1. Self-reported rating of severity also was significantly reduced (p < 0.0001, ES 1.4), with no significant adverse effect on speech naturalness found. A second paper from this research team17 evaluated VSM as part of the maintenance programme following a smooth speech/prolonged speech (PS) intervention. The study (which was judged as at lower risk of bias) compared standard maintenance with VSM over a 4-week post-intervention period. It found that there was no significant difference between standard maintenance and VSM outcomes in terms of percentage of syllables that are stuttered (p = 0.92), self-rated anxiety (p = 0.12) or avoidance (p = 0.69); however, self-reported rating of typical and worst severity were better in the VSM group (p = 0.062 and p = 0.012). Participants in this group rated their satisfaction with fluency as greater (p = 0.043) and quality-of-life scores were higher (p = 0.027).

Cognitive interventions

This category of interventions may have content that includes reduction of tension, anxiety, fear, shame or stress; or a greater acceptance or feeling of control over stuttering; improved self-esteem; or more positive perceptions of own communication and desensitisation to the stutter. The interventions aim to effect change in psychological or psychosocial processes in people who stutter. This type of intervention may be used alone or to support, optimise or prepare for other interventions and may traditionally have been delivered by counsellors or psychologists. However, it is increasingly being perceived as part of a speech and language pathologist’s role, particularly in the UK. The anticipated outcomes may be direct speech gains, psychological well-being gains that lead to improved speech, or alternatively gains which do not aim to change the frequency or severity of the stutter but instead relate to living successfully with stuttering.

Six papers were identified within this intervention typology. There was one paper published in the early 1990s,76 with other articles published 2002–12. 18,30,66,67,93 The work originated from a broad range of countries (Ireland, Australia, the Netherlands, India, Turkey) and all were judged to be at higher risk of bias (Table 9).

Two papers evaluated CBT. 30,93 Reddy et al. 93 presented the study findings as a series of five descriptive case reports only. The article outlines that SSI scores improved between pre and post assessment. It presents a formula for calculating percentage of therapy change but fails to include these data. Reportedly, three clients had clinically significant improvement in anxiety symptoms and dysfunctional cognitions and there was some benefit in self-reported quality of life, but details of this are very limited. An intervention reported by Amster and Klein30 was described as having CBT as the main focus; however, also included stuttering modification treatment for the final 8 out of 12 sessions. The study found a significant decrease (p = 0.035, ES 1.80) in participant self-rating of perfectionism during the early weeks of the treatment to mid-point, which was maintained at 15-week follow-up (no pre- to post data provided). Participants reported improved communication attitudes at the end of the programme and at follow-up (p = 0.017). Speech fluency scores using SSI were mean 24.38 at baseline, mean 11.75 post treatment and mean 13.75 at follow-up (ES 0.74 pre-treatment to mid-treatment and ES 0.51 mid-treatment to post treatment).

Leahy and Collins76 based a 10-session group intervention on personal construct psychology (PCP). Changes in SSI are reported by individual and range from 3 to 31 pre-intervention and from 0 to 10 post intervention for the five participants. However, this evaluation of PCP has a significant flaw as the clients received concurrent individual fluency therapy.

A paper by Kaya and Alladin66 and another by Kaya67 describe the use of hypnosis alone and hypnosis combined with diaphragmatic exercise for people who stutter. Rating of fluency pre and post intervention showed a significant effect (p < 0.000) with informal patient report in the later paper stating that all but four participants were ‘doing well’. Although these papers have reasonably large sample sizes (93 and 59, respectively) the rating scale used for evaluating fluency has considerable limitations.

The Mindfulness-Based Stress Reduction programme as a potential intervention for stuttering was examined by De Veer et al. 18 Although this is one of few studies identified that used a controlled design, the recruitment and allocation process together with only self-report measures resulted in a rating of higher potential for bias. The authors found a significant difference post intervention between intervention and control groups in measures of stress (p < 0.001), anxiety about speech situations (p < 0.01), self-efficacy trust (p < 0.01), locus of control (p < 0.001), coping (p < 0.05) and attitude towards speech situations (p < 0.01). Average ESs were found for self-efficacy beliefs, coping and attitude towards speech situations (d = 0.55, 0.62 and 0.48, respectively). ESs were large for stress, anxiety and locus of control (d = 1.16, 1.07 and 0.76, respectively). There was some maintenance of these positive outcomes at the 4-week follow-up.

Behaviour modification

The greatest number of papers identified that related to a single intervention was the LP. This intervention is based on operant conditioning principles with the content focusing on training parents to provide feedback (verbal contingencies) for stuttered speech and stutter-free speech. The precise mechanism of change whereby verbal contingencies lead to a reduction in frequency of stuttering is unclear and may include neural reorganisation, motoric alterations or changing system demands. In the LP, thresholds of percentage of syllables stuttered and stuttering severity determines progress from the first to the second stage of the intervention.

Twenty-two papers considered aspects of the programme including effectiveness in the short and longer term, predictors of treatment time, predictors of responsiveness, applicability in different countries and components of intervention delivery such as telehealth. 22,24–27,29,42,51,58,65,68,69,77,82,86,88,96,112,113,121–123 These papers compared intervention with no intervention and originate predominantly from Australia. An additional paper54 compared the LP with Demands and Capacities intervention and is therefore considered in detail in the section on papers comparing programmes. This intervention type, as well as having the largest number of papers, also tended to be where the quality was higher with 12 papers assessed as being at lower risk of bias (Table 10).

Behaviour modification programmes such as the LP are used largely with pre-school children. It is important to note that studies evaluating interventions in this population face the challenge of demonstrating not only if the intervention achieves change, but also need to fully consider the possibility of spontaneous remission of stuttering in participants. Although there is some variation in reported rates of spontaneous improvement, the figure is generally recognised as being in the region of 80% of children. 114 The recovery figures relate to a general population but precise figures for spontaneous improvement in clinical populations are currently unknown. The length of time since onset is believed to be a significant influencing factor in whether or not development stuttering resolves. In order to demonstrate clear evidence of effectiveness in populations of young children, interventions need to demonstrate not only evidence of effectiveness but change beyond a level of 80% recovery.

Of the 11 papers focusing primarily on clinical effectiveness of the LP, four reported early data from the 1990s. 29,77,88,123 These studies found positive effects on the percentage of syllables that are stuttered for small groups of participants, and indicated benefits (achievement of < 1.5% of syllables are stuttered) continuing to 12-month follow-up. One of these papers highlighted ethical issues with control group designs for this population. 29 The seven more recent articles were published between 2000 and 2012 and confirm the effectiveness of the LP using larger groups and stronger study designs. Harris et al. 22 found a significant mean reduction in the percentage of syllables that are stuttered of 39% (p < 0.001) pre- to post intervention in the nine children in the LP intervention group. This compared with a reduction of 16% for nine of the children who had not received the intervention and an increase of 6–54% of syllables that are stuttered in four other children in the control group. Owing to the control group design, this paper was able to demonstrate a greater improvement than spontaneous remission alone (although randomisation was quasi- rather than fully randomised).

The papers by Jones et al. 24,25 present data from a RCT with immediate, 12-month and up to 5-year follow-up. The first paper reported a reduction of 2.3% of syllables that are stuttered at 9 months and the second paper reported that 16 out of 19 participants who had completed the intervention and could be contacted had 0–1.1% of syllables that are stuttered at 5-year follow-up. Three participants had relapsed to pre-intervention levels; however, the reduction between pre-intervention and follow-up remained significant (p < 0.0001). Parental satisfaction was high and none of the children had received treatment other than the LP. Of the few participants in the control group who could be contacted, five out of the eight were reported to have recovered spontaneously.

Four papers published between 2008 and 2012 add further strength to the evidence of effectiveness of the LP. Lattermann et al. 26 evaluated use of the programme in Germany using a randomised design and a sample of 46 children. The intervention group decreased the percentage of syllables that are stuttered by 6.9% at home measurement compared with the comparator waiting list group reduction of 1.6% of syllables that are stuttered at 16 weeks post intervention. The in-clinic measures showed a similar reduction of 6.8% of syllables that are stuttered for the intervention group compared with 3.6% of syllables that are stuttered in the comparator group at 16 weeks post intervention, with a significant effect (p = 0.003 home and p = 0.025 clinic). The reduction in the percentage of syllables that are stuttered was not at the expense of a reduction in rate of speech.

Femrell et al. 51 reported outcomes at 2 years following intervention with the LP in Sweden. Eight out of the 10 participants completed the programme with a significant reduction in the mean percentage of syllables that are stuttered from 7.6% to 0.1%, a large ES of d = 2.9. The two dropouts withdrew early as parents were satisfied with the benefits achieved. Parents had been offered the choice of receiving the LP or an alternative intervention – all had chosen the LP. Koushik et al. 69 investigated use of the programme with older children (aged 6–10 years). Mean percentage of syllables that are stuttered at baseline was 9.2% and 1.9% of syllables are stuttered at follow-up, with no adverse effect on speech rate.

O’Brian et al. 84 explored the use of the LP in community clinics, rather than specialist centres. The study found a mean parental rating of severity at baseline of 5.2 and at 9-month follow-up parental rating of severity was mean 2.1. At 9 months, the mean percentage of syllables that are stuttered was 1.7% (no baseline provided) with a range of 0.1% to 13% of syllables stuttered and 47% of the children had a percentage of syllables that are stuttered below 1%. Some individual variation was highlighted, with 6 out of the 37 completing stage 1 reportedly having high severity ratings of 5–7. In addition to clinical effectiveness, this paper examined factors contributing to outcome. It reported that the clinician having a high level of specialist training was important in achieving optimal outcomes and that more severe stuttering was associated with longer intervention duration and higher dosage.

Nine further papers relating to the LP consider implementation issues and predictors of outcomes. Miller and Guitar82 replicated findings of effectiveness (an 86% reduction in stuttering severity measured by SSI p < 0.001, ES 3.7) reported in other papers when the programme was delivered by less experienced but specifically trained clinicians assisted by student clinicians. Children with more severe stuttering pre-intervention (≥ 5% of syllables stuttered) required a longer treatment time. Pre-treatment severity also predicted number of clinic sessions received in the Koushik et al. 121 and Rousseau et al. 96 studies. In Koushik et al. ,69 there was no association between sex, age or onset to treatment time and outcome; however, there was a seemingly counter-intuitive association between more frequent attendance and longer treatment time. Two further papers65,68 echoed the association between severity and time needed for treatment (OR 3.5 for more severe stutter to take longer to treat; p < 0.001). Eleven sessions was the typical length of treatment to complete stage 1. Although the earlier paper65 found a lack of association between onset to treatment time and outcome, Kingston et al. 68 (which combined data with the Jones et al. 65 study) detected an association between children stuttering for longer before treatment and reduced treatment time (OR 0.52).

Papers by Woods et al. ,113 Onslow et al. 123 and Bonelli et al. 42 explored the outcomes and impact of the LP. The first of these113 reported no evidence of an adverse effect on child behaviour, child mental health or parent–child relationship over the course of intervention for eight children and their mothers. The second42 found a positive effect on maternal speech rate following the programme for nine mothers (who had taken part in earlier studies). Onslow et al. 123 concluded that there was no evidence of an adverse impact on speech timing or language function in eight children (two of whom were in the Bonelli study42 and six in earlier studies77,122). Harrison et al. 58 aimed to evaluate which components of the LP may be the more important factors underpinning outcome. The authors evaluated (1) 4 weeks of parental requests to self-correct (verbal contingencies) compared with 4 weeks with no parental correction and (2) 4 weeks of parents completing severity ratings compared with 4 weeks of no rating. They concluded that parental verbal contingencies were likely to be the active element rather than completion of rating checklists.

The final two papers examining the LP investigated the potential for delivery of the programme via telehealth. Lewis et al. 27 concluded that telehealth delivery was effective (73% reduction in stuttering compared with a no intervention group; p = 0.02); however, required additional clinician input (costing around three times more than the standard version). A later paper112 in a small study with high dropout rates confirmed the clinical effectiveness but need for greater number of consultations for telehealth delivery of the programme.

Other interventions that we categorised as behaviour modification focused on changing behaviours within the family, predominantly parent behaviour and parent–child interaction. Four papers were identified which evaluated these interventions in children, all of which were rated as being at higher risk of bias. 78–81 As with the LP, they were primarily targeted at pre-school children and, thus, need to consider spontaneous recovery within their assessment of outcomes. Two additional papers in this category evaluated behaviour modification programmes with adults (Table 11). 21,23

Two papers from the UK by Millard80,81 evaluated parent–child interaction therapy. This intervention combines helping parents to manage their child’s stuttering through parent-identified interaction targets (such as reducing their rate of speech or complexity of language), with family strategies to develop confidence. In the earlier paper,80 which report data by individual participant, six children were followed up for a 12-month period. Three of these had reduced their stuttering severity on a 0–7 scale from 2, 3 and 5 to zero (normal speech), one had reduced from 2 to 1, one was unchanged and one reportedly did not reduce with this intervention, but reduced from 5 to 2 with a period of direct intervention. The later study81 aimed to use a randomised design but was forced to remove this comparator condition mid-way through. Data are reported by individual participant using cumulative sum control charts. Of the six children in the intervention group, four reportedly showed systematic reductions in stuttering frequency from baseline to follow-up at 12 months, which may be attributable to the intervention. One out of the four control group children showed a systematic reduction. From chart data, parental ratings of child fluency and confidence in managing stuttering appeared to increase.

Yaruss et al. 115 report a family-focused treatment targeting parent communication modification and parent and child understanding and acceptance of stuttering. Following the training there was a significant reduction in the 17 children’s dysfluencies (p < 0.001) as rated by the clinician. Parental rating scale data also indicated improved fluency and overall satisfaction with the treatment. The parent education component was rated as the most helpful element. Eleven of the children were discharged from therapy without requiring direct child intervention. Another study examining parent understanding and acceptance was reported by Lutz. 78 This weaker-quality paper reported that following a weekend parent workshop, 92% of participants rated themselves as having changed their attitude towards stuttering.

The Rustin programme was evaluated by Mallard. 79 This is a family-oriented intervention which includes a range of elements encompassing speech skills, transfer skills and social skills, which is delivered via children groups and parent groups. The therapy emphasises that families need to find the most appropriate intervention methods for them, with assistance from the clinician. The children in this study were school aged (5–12 years) rather than pre-school aged and the paper reports that 23 out of the 28 families (82%) did not receive any further intervention following the programme. The authors noted that the areas of the programme rated as most important by parents ‘had nothing to do with speech modification’, and instead were ‘letting the child take responsibility’, ‘family discussion’ and ‘listening’.

Although all the above interventions are used with children, an additional intervention that we categorised as behaviour modification is used with adults who stutter. Self-imposed time-out treatment is, like the LP, based on an operant conditioning approach. This intervention was evaluated in two papers that met our inclusion criteria, one rated as lower and one as higher risk of bias. This treatment involves participants learning to modify their behaviour by pausing for a moment after a stuttering episode. Hewat et al. 23 found individual diversity in response to the intervention. The mean reduction in the percentage of syllables that are stuttered from pre-treatment to post stage 1 was 53.6%. More than half (from figure total 13 of 22 [sic]) the participants reduced their stuttering frequency by more than 50%. Six participants reduced by 50–60%, three participants by 60–70% and four had an 80–90% reduction (numbers are approximate as taken from figure). Speech naturalness was judged as being poorer than non-stutterers post intervention but compared favourably with people who had completed PS treatment. Participants reported general satisfaction with the intervention.

The second paper assessing time-out19 found a significant reduction in the percentage of syllables that are stuttered between baseline and post treatment (mean 5.8% vs. 3.9% syllables stuttered; p < 0.007). A control group increased their stuttering in same time frame (from baseline 4.9% to 6.4% syllables stuttered; p < 0.007). There was no adverse effect on speech rate, with the intervention group increasing syllables per minute (SPM) post intervention. There was evidence of an association between stuttering severity and outcome (more severe responded better), and amount of previous therapy and outcome (more previous therapy responded better).

Speech motor interventions

Eighteen papers evaluated interventions that we classified in our typology as speech motor interventions. 31,37,38,45,46,49,53,62–64,84,85,89,90,104,106,107,114 The content of these interventions is focused on the mechanisms of speech production (breathing, vocal fold vibration, articulation of sounds) with reduction in the severity or frequency of stuttering achieved by altering speech motor patterns. People who stutter may be taught to change their speech pattern, for example by prolonging sounds, reducing speech rate or making articulation more soft or smooth. These interventions are referred to variously in the literature as ‘behavioural treatments’, ‘talk more fluently approaches’, ‘speech restructuring’ and ‘fluency shaping’. In order to be clear within our typology, we have labelled them as ‘speech motor’ rather than ‘speech behaviour’, in order to avoid confusion between these therapies and interventions targeting parent–child behaviour. These interventions typically include a clinician modelling the desired pattern and teaching the participant to use it. As the approach entails changing a participant’s usual pattern of speech, an important aspect to consider when evaluating speech motor interventions is not only the degree to which the therapy reduces the frequency or severity of stuttering, but also whether the speech produced using the changed motor pattern is acceptable to the speaker (and listeners) or whether it sounds slow and unnatural.

Although these interventions were given various labels, the largest group were described as consisting of teaching PS. This included the Camperdown Program, which is based on control of stuttering using PS. Seven papers from a team at the University of Sydney, NSW, Australia, and La Trobe University, VA, Australia, outline results from evaluation of PS treatment. These papers were published between 1992 and 2012, with four (mostly older papers) graded as being at higher risk of bias,84,85,89,90 and three graded as being at lower risk of bias. 38,46,47 Table 12 shows a summary of these studies.

The lower-quality papers reported improvement in the percentage of syllables that are stuttered following intervention and at up to 12-month follow-up. O’Brian et al. 84 for example found pre-treatment mean 7.9% of syllables stuttered reduced to 0.4% of syllables stuttered at 12-months maintenance and Onslow et al. 90 found the percentage of syllables that are stuttered generally at, or near, zero for 9 out of the 12 clients (the other three scored > 1%), and a mean 82% reduction in stuttering frequency from baseline to immediate follow-up and 74% reduction at 6 months. 85 Evaluation of mean naturalness scores indicated that post intervention there was no significant impact on naturalness;89,90 however, one paper noted that people who stutter could be distinguished from non-stuttering speakers (mean 4.5 vs. matched control participants mean 3.6; p = 0.025) although the difference was less than one naturalness scale value. 84 Speech rate also did not appear to be adversely affected, for example all participants increased speech rate with the group mean increasing from 184 SPM to 228 SPM. 85

The three papers judged as being at lower risk of bias, similarly reported positive effects of PS intervention. Block et al. 38 reported a pre-treatment mean percentage of syllables that are stuttered of 4.9% (SD 4.4%). Levels of stuttering reduced to a mean of 0.9% (SD 1.4%) of syllables stuttered immediately post treatment and 1.5% (SD 2.2%) of syllables stuttered at 3-month follow-up. At 12 months, the mean percentage of syllables that are stuttered was 2.6% (significant difference from baseline; p = 0.04) and at 3.5–5 years, during a surprise telephone call, the mean percentage of syllables that are stuttered was 1.6%. Carey et al. 45 echoed positive outcomes using a telehealth delivery and conventional delivery of the Camperdown Program. Cocomazzo et al. 46 used supervised student clinicians to deliver the intervention and found that similar outcomes could be achieved to that obtained by qualified clinicians (pre-treatment percentage of syllables that are stuttered 5.7%, immediate post treatment 1.0% of syllables stuttered, 12-month follow-up 2.4% of syllables stuttered, ES 0.61–0.75). Speech naturalness scores echoed the earlier work38 in finding that participants who completed the treatment had scores averaging one scale point below (less natural) than non-stuttering speakers. Variation in effect on naturalness was described by Cocomazzo et al. 46

Block et al. 38 examined possible predictors of successful outcomes and concluded that only baseline stuttering severity and short-term response to intervention predicted longer-term outcomes. Age, sex, perceived locus of control, attitude to communication or previous treatment did not predict long-term outcome. The authors highlighted that 46% of variance between participants at long-term follow-up was unaccounted for.

Another paper that evaluated speech motor programmes in adults examined a smooth speech intensive treatment. 37 The study found the reduction of the percentage of syllables that are stuttered following treatment was statistically significant (pre-treatment 5.4% of syllables stuttered and post treatment 1.8% of syllables stuttered, large ES 0.86) and at 3.5- to 5-year follow-up the mean stuttering rate was 1.6% of syllables stuttered. A paper from the Netherlands53 found an improvement from 27.7% of syllables stuttered pre-intervention to 5.8% of syllables stuttered post intervention and change from baseline but considerable relapse to 16.3% of syllables stuttered at 6-month follow-up. There was no impact on speech rate and some positive effect on rating of speech distortion; however, dynamics/prosody rating was no nearer to non-stuttering speakers post intervention than it has been before the therapy. Iverach et al. 64 examined whether or not the presence of a mental health disorder had an impact on outcomes following speech motor intervention. The authors found that stuttering frequency and situation avoidance were significantly worse for participants who had a mental health disorder.

Two papers by Ingham et al. 62,63 judged to be at higher risk of bias examined the use of Modified Phonation Intervals. Ingham et al. 63 described five adults as achieving stutter-free and natural sounding speech immediately and at 12-month follow-up after completing the intervention. The later paper62 focused on examining brain activity as a potential predictor of outcome following modified phonation intervals or PS intervention, and reports some positive outcome data (a pre-treatment mean of 7.1% of syllables stuttered and end of treatment mean percentage of syllables that are stuttered of 1.0%).

Speech motor interventions are not only used with adolescent and adult populations but also with children. von Gudenberg106 and von Gudenberg et al. 107 evaluated Kasseler Stuttering Therapy, Yairi and Ambrose114 describe slow speech therapy outcomes and Druce et al. 49 describe their intervention as most closely approximating the Gradual Increase in Length and Complexity of an Utterance (GILCU) model. This was the only paper evaluating speech motor interventions with children that was rated as being of lower risk of bias49 but all papers reported positive outcomes. The von Gudenberg et al. 107 paper from Germany found large ESs for their treatment comparing baseline to 1-year follow-up (d = 0.96 for 9- to 13-year-olds, and d = 0.88 for 14- to 19-year-olds). The other paper evaluating the Kassel smooth speech and PS treatment106 reported no adverse effect on speech naturalness or speech rate and improved self-perception of their speech among participants. Druce et al. 49 also reported positive outcomes. From pre-intervention to after the intensive week intervention the mean percentage of syllables that are stuttered for the group decreased by 7.6% to 1.75% (p = 0.0015).

The Yairi and Ambrose paper114 differs from the others in the group by using a natural history approach to compare pre-school children who received an intervention with no-intervention controls, rather than evaluating the clinical effectiveness of a specific intervention. The study highlights the importance of considering natural recovery in this population, as it found that although the intervention group reduced their level of stuttering, there was no significant difference between this group and untreated children over time. Both had a downwards trend in dysfluency and there was no significant difference between them (p = 0.4).

The final papers in this group examined interventions for pre-school and school-aged children termed syllable timed speech (STS). The paper judged at lower risk of bias104 evaluated STS and reported a mean stuttering reduction of 96% in beyond-clinic conversations from pre-treatment 6% of syllables that are stuttered to 12-month follow-up 0.2% of syllables that are stuttered (large ES 1.8). In another evaluation of STS, Andrews et al. 31 found the group mean percentage of syllables that are stuttered reduced from 14.4% at baseline to 6.7% at follow-up (p = 0.015, medium ES 0.7). Data on self-reported severity, situation avoidance and quality of life confirmed these positive outcomes, but the authors noted considerable individual variation in response to the intervention.

Speech motor combined with cognitive elements

As mentioned above (see Cognitive interventions), cognitive interventions may be used as an intervention type in isolation, or alternatively may form part of a programme. Eighteen papers reported interventions that combined speech motor therapy with elements of cognitive interventions (Table 13). In contrast with the cognitive interventions only category, in which all papers were considered to be at higher risk of bias, one-third of these papers combining cognitive with speech motor elements were judged to be at lower risk of bias.

The Comprehensive Stuttering Program (CSP) for adolescents and adults was evaluated in six papers,41,61,71–74 with all but one of these from a research team in Canada. This intervention incorporates speech motor techniques with cognitive strategies to have an impact on emotional and attitudinal aspects in addition to speech fluency. Of the three papers published in the 1990s,41,71,72 only the later paper72 was rated as being at lower risk of bias. All papers reported a substantial reduction in client percentage of syllables that are stuttered following intervention. The later paper72 reported that four participants improved stuttering by 55–99% following the CSP, mean percentage of syllables that are stuttered at baseline was 61.3% and 12.5% at immediate follow-up. Langevin and Boberg (1993)71 reported a pre-treatment mean of 15.3% of syllables that are stuttered during a telephone call, 0.8% post treatment, and 2.4% of syllables that are stuttered at a 12- to 14-month follow-up telephone call. Boberg and Kully41 found pre- to immediate post-treatment mean percentage of syllables that are stuttered decreased from 19.59% to 1.29% for the adult group and 14.32% to 1.75% for the adolescent group. As with the Langevin and Boberg paper,71 this study indicated some fading of effect, as percentage of syllables that are stuttered had increased from 1.29% immediately post intervention to 4.27% at 4 months and 6.03% at 12 months for the adult group, and from 1.75% immediately post treatment to 3.65% at 4 months and 3.89% at 12 months for the adolescent group. The authors found no adverse effect on speech fluency following the CSP, with the mean SPM increasing from 126.5 pre-treatment to 140.7 post treatment. The Langevin and Boberg71 paper is interesting in that it reported data for people who exhibit cluttering in addition to stuttering, as a separate group to people who stutter. They noted that adults who clutter respond more poorly to intervention.

The three more-recent papers all had stronger designs and were judged to be at lower risk of bias. One paper reports long-term outcomes in Canada,74 another considers predictors of outcomes in the Netherlands61 and the third compares the findings across Canadian and Dutch populations. 73 Langevin et al. 74 present 5-year follow-up data from 18 adults following treatment with CSP (including some participants that were reported in the 1993 and 2006 studies). Since the earlier study,72 eight individuals had attended refresher sessions and 10 had not. Pre-intervention mean percentage of syllables that are stuttered was 15.86% and immediate post-intervention percentage of syllables that are stuttered was 0.90% (pre- to post intervention; p < 0.001, ES d = –2.07), with 1-year follow-up mean of 3.59% of syllables that are stuttered. At subsequent yearly time points, the percentage of syllables that are stuttered remained reasonably stable (4.38%, 3.81%, 3.76%, 4.98%) with pre-intervention to 5-year follow-up reduction significantly different (p = 0.02, ES d = –1.16). Other measures [S24, Self-Efficacy Scaling by Adult Stutterers (SESAS), PSI] also indicated self-report benefits for participants at 2 years post intervention and SPM increased following treatment. Langevin et al. 73 compared data sets from Canada and the Netherlands, with ESs of d = 0.52 (Dutch), d = 0.86 Canada and d = 0.69 (pooled). For the Dutch group, 71% were categorised as maintaining clinically significant reductions at 2 years, and 86% in the Canadian group. Both groups of participants had mean naturalness ratings that were within the range of mean ratings reported for non-stutterers (range 2.3–3.6).

Individual variation in patterns reported in studies of the CSP was investigated by Huinck et al. 61 They found that those with the most severe stutter pre-intervention had the most immediate gain, but tended to be more likely to regress. Severity of stuttering did not predict severity of negative emotions, with people having more negative emotions tending to rate their stutter as worse than it actually was.

In three papers39,70,75 authors described their intervention for adults only or adolescents and adults as being based on stuttering modification techniques used by Van Riper. 127 The Successful Stuttering Management Program includes confrontation of stuttering (targeting attitudes and perceptions), stuttering modification techniques (prolongation, cancellation and pull outs) and maintenance. Blomgren et al. ,39 in a study judged to be at lower risk of bias, found statistically significant improvements at 6 months post treatment on client-reported perceptions (the Avoidance and Expectancy subscales of the PSI; p < 0.001) and two specific affective functioning measures (the Psychic and Somatic Anxiety subscales of the Multicomponent Anxiety Inventory IV (MCAI-IV); p = 0.078 and 0.036, respectively). However, statistically significant reductions were not evident on objective measures of dysfluency for the adult participants (during monologue, mean 17.8% pre-intervention to 11.8% post intervention, and 13.8% at 6 months post intervention). Another intervention drawing on Van Riper’s methods127 for therapy with adults was outlined in Lawson et al. 75 This method was described as combining block modification, avoidance reduction and elements of PCP. The study found positive changes in avoidance on PSI but no change in percentage of syllables that are stuttered, struggle or expectancy. The positive effect on avoidance was unchanged at 1-month follow-up.

Laiho and Klippi70 evaluated an intervention drawing on Van Riper’s methods127 with children and adolescents. The intervention had a positive effect on stuttering severity (a reduction in percentage of syllables that are stuttered) for 14 out of the 21 participants (mean percentage of syllables that are stuttered pre-intervention was 4.45% and post intervention was 2.7%: a 38% improvement; p = 0.001). The amount of avoidance behaviour also reduced for 13 participants (p = 0.001) and positive feedback was received from participants. Rosenberger et al. 95 combined stuttering modification with social interaction activities and awareness tasks for children who stutter. Positive effects were found for stuttering rate (p < 0.001) and anxiety (p < 0.025).

Other papers reporting outcomes from programmes for children which combined speech motor and cognitive elements include Berkowitz et al. ,34 Smits-Bandura and Yovetich99 and Baumeister et al. 32 The Smits-Bandura and Yovetich99 programme achieved some reduction in behavioural and attitudinal stuttering symptoms; however, it was reported that 90% of the participants required further therapy after 2 months. Berkowitz et al. 34 outlined findings following intervention with the Cooper Personalised Fluency Control Therapy Program. Although there appeared to be some benefit, the results are reported as scores on individual tests and general description only. Baumeister et al. 32 found a significant reduction in stutter frequency following an intensive summer camp encompassing speech, cognitive and social elements. The data are limited by analysis of different groups of participants at different time points, but indicates a reduction from 22.2% of syllables that are stuttered to 9.5% of syllables that are stuttered (ES 1.29).

Other papers reporting outcomes for adults from speech motor and cognitive interventions include Nilsen and Ramberg,83 Ward,110 Stewart100 and Irani et al. 116 All were rated as being at higher risk of bias. Irani116 is notable as the only study we identified that used a mixed-methods study design. The paper has therefore been included in both the review of effectiveness and the qualitative review. The intervention combined cognitive and speech motor elements including CBT in an intensive programme. The percentage of syllables that are stuttered pre-intervention to post intervention during conversation was significantly improved with a large ES of 1.12. S24 scores (ES 1.19) and locus of control of behaviour (LCB) assessment (ES 0.75) also indicated benefits pre- to post intervention. Participants were interviewed at a follow-up interval of between 2 and 6 years post intervention, with improvement compared with baseline sustained (ES 1.25–1.97; p < 0.07). However, the table of participants details that three out of the seven participants received more than one course of intervention.

Ward110 evaluated semi-intensive fluency therapy (SIFT), which is described as similar to CSP but with identification, prolongation and transfer phases. Pre-intervention percentage of syllables that are stuttered for the group during conversation was 10.2% and post-intervention percentage of syllables that are stuttered was 3.3%. Stewart100 examined attitude change during therapy and maintenance. At baseline, the mean percentage of words stuttered was 30.6% (SD 28.28%). After attitude change sessions, the percentage of words stuttered group mean was 30.7% (SD 34.5%). After 1 year, the group mean was 12.6% (SD 25.78%) and after 2 years the group mean was 19.7% (SD 18.9%). The author noted that the specific attitude-change sessions did not seem to result in significant changes, but change was apparent in most of the attitude measures following the technique sessions. During transfer and maintenance, the group maintained speech gains but a small number of participants had poor maintenance. A paper from Sweden83 used independent listener and therapist ratings of change. Overall, 12 out of the 13 participants were rated as having improved on at least one aspect measured. The most recently published paper in this final group of studies is Allen,117 a paper from the UK, which examined e-mail as a component of a speech modification and counselling intervention. The limited evaluation data outline that out of the 16 clients who used e-mail as part of therapy, 11 were discharged (two owing to non-response) and five clients remained on the caseload.

The final paper in this group assessed the effectiveness of Acceptance and Commitment Therapy (ACT). 33 The intervention was carried out weekly over 8 weeks and included mindfulness skills in the programme. Although this study used a before-and-after design with no comparator group and a large number of self-report measures, it included speech data and exhibited rigor in collection and analysis of data and, therefore, was rated as comparatively at lower risk of bias. Results from this study showed statistically significant gains across all measures from pre- to post treatment and at 3-month follow-up. Percentage stuttered syllables reduced from pre-intervention mean 6.42% to post-intervention mean 1.39% and mean 1.77% at follow-up (p < 0.001). Psychological measures such as OASES also improved significantly post intervention and at follow-up (p < 0.001).

Multiple elements

Eleven papers described interventions that included multiple components across our typology of interventions, or were papers that evaluated a range of interventions (Table 14).

Three treatment programmes included EMG feedback, one for children and two for adolescent clients. Hancock and Craig57 and Craig et al. 47 examined a retreatment programme for adolescents who were experiencing difficulty maintaining fluency following intervention. The therapy included EMG, smooth speech, relaxation, cognitive and self-management components. The follow-up level of percentage of syllables that are stuttered at 12 months was no different for the retreatment intervention compared with the initial intervention. However, the retreatment intervention resulted in significantly lower percentage of syllables that are stuttered at 2-year follow-up than the initial intervention 2-year follow-up. The SPM scores were also significantly better for retreatment at 2-year follow-up than initial treatment 2-year follow-up. Narrative in the Craig et al. 47 paper describes individual difference in response, with two participants showing immediate improvement after the relapse programme; however, they had relapsed to > 5% of syllables that are stuttered at 2-year follow-up. Two other participants reportedly improved quickly and gains were maintained (‘well below 5% of syllables that are stuttered’) at 2 years. The final two participants reportedly improved more slowly but at 2 years remained ‘well below’ 5% of syllables that are stuttered. Hasbrouk59 also described a treatment programme including EMG combined with airflow training, relaxation and desensitisation used with adult military service personnel. The mean number of stutters for the 151 participants reduced from 5.34% to 0.18% with all reaching the criterion level of < 1% stuttered words. The author noted that the programme was less effective for those with more severe stuttering at baseline.

The intervention reported by Blood40 involved motor speech changes assisted by a biofeedback computer program together with POWER, a relapse management prevention approach targeting self-efficacy and cognitive–behaviour modification. At the end of phase 1, all participants had reduced stuttering to the criterion level of < 3% of syllables that are stuttered. Two increased percentage of syllables that are stuttered to > 3% during the second and third phases; however, did not relapse to pre-treatment levels. The feeling and thinking scales all showed positive changes, which were maintained at 6 and 12 months.

Four other papers included regulation of airflow in the intervention components. These papers from North America report intervention with 5- to 11-year-old children. The interventions included regulated breathing, awareness training, social support, easy speech and relaxation. Positive outcomes following intervention were reported for the majority of participants across the studies. Elliott et al. 50 found four out of the five children reduced stuttering to < 3% stuttered words, Gagnon and Ladouceur55 described a similar reduction with gains retained at 6-month follow-up. Wagaman et al. 108 reported that all eight children reduced the proportion of stuttered words to < 3% and that parents rated the intervention as acceptable. A paper reporting longer follow-up data from this study109 found that for five out of seven participants, the follow-up mean percentage of syllables that are stuttered at 3- to 5-year follow-up was lower than at 1-year follow-up. For the other two participants, the mean percentage of syllables that are stuttered had increased from 1-year follow-up level but the score remained well below their pre-intervention stuttering levels.

Pape-Neumann91 evaluated interventions which were chosen by clinicians rather than examining a particular therapy type. For any intervention delivered to children, data indicated a reduction in stuttering frequency of ES 0.63 post intervention. For adolescents and adults, the ES across any intervention was 0.77. A positive impact on avoidance of communication, attitude towards communication, self-judgement of stuttering in social situations and everyday life was also recorded across the therapies. Sicotte et al. 98 examined the feasibility and application of telemedicine across an unspecified typical therapy for six children and adults who stutter. Data are limited but all participants improved fluency with some benefits retained at the 6-month follow-up.

Papers comparing interventions

Our final typology contains papers that had the purpose of directly comparing interventions with each other. We identified eight papers that compared interventions with one another (rather than having no comparator, or comparing an intervention with no intervention). These papers were generally of reasonable quality, with only two considered to be at higher risk of bias (Table 15).

Franken et al. ,54 in a paper judged as being at lower risk of bias, compared the LP with Demands and Capacities Model (DCM) treatment for pre-school children. For the LP, the mean stuttering frequency within an audio-recorded sample decreased from 7.2% (SD 2.0%) at baseline to 3.7% (SD 2.1%) post intervention. For the DCM treatment, the means decreased from 7.9% (SD 7.1%) at baseline to 3.1% (SD 2.1%) post intervention. Stuttering severity was rated on a scale by clinician and parent with a significant effect from pre-intervention to post intervention (p < 0.01) for both interventions with no significant difference between them (p > 0.10).

Menzies et al. ,28 in another higher-quality study, compared speech motor intervention alone with speech motor combined with CBT. Post-treatment percentage of syllables that are stuttered at follow-up was around half that at baseline. The authors found no difference in percentage of syllables that are stuttered between the two groups, with the additional CBT treatment having no additional impact on the stuttering reduction than speech restructuring treatment alone. Although not affecting speech outcome, the group who received the CBT showed greater improvement on measures of anxiety and avoidance.

Bioresonance therapy was compared with standard speech therapy in a study from Germany. 111 The groups received one therapy for 10 sessions and then swapped to the other therapy for the second 10 sessions. There was some improvement in fluency during the first 4 months of therapy, but no further improvement in the second therapy phase for which intervention programmes changed. Data were limited and there was considerable variation in individual response to the intervention, meaning that was not possible to conclude whether or not bioresonance therapy was more effective.

The method of gradually increasing length and complexity of utterances was compared with DAF by Ryan and Van Kirk. 97 The study found that, although both interventions achieved a significant reduction in the percentage of syllables that are stuttered (p < 0.01), the GILCU programme was slightly superior in terms of generalisation of fluency. Riley and Ingham94 compared the effect of speech motor training (SMT) (emphasis on speech motor skills) with extended length of utterance (ELU) intervention (response-contingent feedback without direct SMT), specifically on vowel duration measures and stuttering frequency. Across both interventions there was a median decrease in the percentage of syllables that are stuttered of 3.19% (41% reduction), with 37% reduction for SMT and 64% reduction for ELU (the ELU intervention had a significantly greater effect; p = 0.04).

The Craig et al. 16 study mentioned earlier (see Feedback and technology interventions) compared EMG with intensive smooth speech and home-based speech for children and adolescents. All the therapies included reward and response contingencies, overcorrection, transfer, maintenance and self-monitoring. The study found that the percentage of syllables that are stuttered was significantly reduced baseline to post intervention and at 3-month and 12-month follow-up. The three treatments were found to be equally effective (p = 0). Two further papers20,21 examined longer-term outcomes and possible predictors of relapse following these interventions. At 4- to 6-year follow-up there continued to be no significant difference between the interventions in terms of effectiveness. An association was suggested between having a more severe stutter pre-intervention and being less anxious immediately post therapy, with having a higher level of stuttering in the long term.

Measurement of effectiveness

The most significant challenge in comparing clinical effectiveness between different studies and interventions is the vast range of outcome measures used to evaluate change following an intervention. Outcomes measured include those relating to the frequency or severity of stuttering (number or percentage of words stuttered, number or percentage of syllables stuttered, rating of stuttering severity, number of stuttering events), perceived self-efficacy/control/esteem, anxiety/stress/depression level, self-perceptions of or attitude to speech/stutter, perception of self/others as being a stutterer, avoidance of words/situations, parent verbal interaction, rate of speech and perceived naturalness. The literature used three main strategies to evaluate the effect on these outcomes: first, by comparing percentage change pre-intervention to post intervention; second, by reporting level of frequency at baseline and again post intervention, and then using statistical means to examine the difference; and third, by descriptively comparing the severity level or need for further intervention pre- to post intervention.

Much of the literature reports the percentage of reduction in stuttered speech pre- to post intervention (e.g. baseline and follow-up assessment of percentage of syllables or words that are stuttered). However, these ‘degree of change’ measures have a significant limitation in that baseline stuttering severity will influence how substantial any positive change can be. An examination of baseline levels of stuttering among participants across studies reveals a high level of variation in the fluency of participants prior to the intervention, between studies and, importantly, within a study. For example, Lincoln et al. 77 reported a baseline mean among participants ranging from 5% of syllables that are stuttered to 18.9% of syllables that are stuttered, whereas Langevin and Boberg71 found a baseline stuttering rate ranging from 3.6% of syllables that are stuttered to 9.4% of syllables that are stuttered. Participants in the Rousseau et al. 96 study appeared to have a low baseline of 3% of syllables that are stuttered. Many studies highlighted that there was considerable individual variation in outcome31,38,52 with some linking this to baseline level of severity (e.g. O’Brian et al. 86).

Although caution is thus required when comparing reports of positive percentage reductions pre- to post intervention between papers with participants who have differing baseline stuttering levels, analysis of reduction of the percentage of syllables that are stuttered within each intervention typology reveals evidence of reduction across intervention approaches. Within the ‘feedback and technology’ group, authors reported percentage reductions in syllables stuttered of between 3% and 87%. In the ‘behaviour modification’ typology, percentage reductions in syllables stuttered ranged from 69% of syllables that are stuttered to 97% for the LP and 53% for the one paper23 reporting this measure using other interventions. In the ‘speech motor’ group, reductions in percentage of syllables that are stuttered were reported varying from 22% to 96%. For ‘speech motor plus cognitive’, the one paper70 using this measure found a 38% reduction in percentage of syllables that are stuttered. Among the ‘multiple components’ group, reductions of 52% and 89% were described and a 36.5% and 63.5% improvement in a comparison paper evaluating ELU and SMT interventions, respectively.

In addition to the requirement to consider baseline stuttering levels when evaluating these papers, it is also important to consider the impact of any change for the person who stutters. This requires consideration of not only change but clinical (or personal) impact of the improvement. It has been proposed that in order to be clinically significant, an intervention should result in a 50% reduction in stuttering. 93 Examining the set of papers (all 111 included studies) reporting percentage change in syllables stuttered with this criterion reveals that six out of the eight ‘feedback and technology’ group, all five of the Lidcombe papers, one non-LP behaviour modification paper, five of the six ‘speech motor’ papers and both of the multiple component interventions reached this level, again confirming that a range of intervention approaches identified could result in clinically significant improvement.

The second method for evaluating change was to report the level of stuttering frequency at baseline and again post intervention. These papers reported effectiveness in terms of p-values or ESs rather than percentage change. 24,30–32,37,48,51,73,74,82,91,104,106,116 In the ‘feedback and technology’ group, an ES of 0.14 was reported for stuttering (during monologue only) in one paper,119 and an ES of 1.1 for reduction of stuttering frequency in another48 and statistically significant changes were reported in nine papers. 4,35,52,56,92,118,120,124,125 In the ‘cognitive’ group, an ES of 0.74 for reduction in stuttering was found in one study30 and three papers reported statistically significant differences. 18,30,66 In the behaviour modification group ESs of 2.9, 2.3 and 2.3 are described,24,51,82 and eight papers provided evidence of statistically significant effects. 22,24–26,51,53,82,84 Speech motor and speech motor plus cognitive approach papers reported ESs of 0.70, 0.86, 1.80, 0.96, 0.88, 1.29, 1.12, 6.86 and 14.9630–32,37,73,74,104,106,116 and seven reported statistically significant effects. 32,39,70,73–75,95 One multiple component paper reported an ES of 0.6391 and three comparison papers described significant effects for each of the interventions they evaluated. 16,54,97 This set of findings thus supports the conclusion that a range of interventions may be effective for people who stutter. As with the percentage change evaluations described above, statistical significance may differ from clinical (or personal) significance and mask individual variation in outcomes. Statistical significance is also heavily dependent on sample size as very small effects can be statistically significant with a large sample, while relatively large effects may not be statistically significant with a smaller sample.

The third approach to evaluating outcomes considered level of stuttering before and again after an intervention, or whether or not further intervention was required. Although there is some debate regarding what is a ‘good outcome’ in terms of the level of stuttering following an intervention, many studies use a ≤ 3% of syllables that are stuttered level as being an acceptable degree of dysfluency and thus may be a target for interventions to achieve. 55 Four papers66,67,76,93 in the ‘cognitive’ group used severity scales to evaluate difference (two a scale developed by the author66,67 and two the SSI76,93), these studies found positive outcomes. The papers reporting the LP often included data from severity rating scales and the programme uses threshold levels of percentage of syllables that are stuttered in order for participants to move through the intervention stages. Four non-LP ‘behaviour modification’ papers78,80,81,115 reported positive outcomes in terms of parent report, stuttering severity or need for further therapy post intervention. Five ‘speech motor’ papers38,46,84,85,90 report reduced levels of percentage of syllables that are stuttered post therapy (to 0.9%, 1.6%, 1%, 0.4%, near 0%), three41,71,72 ‘speech motor plus cognitive’ (to 1.29%, 0.53%, 0.1–3.8%), six40,50,55,59,108,109 multiple component papers (to < 3% in four papers,40,50,55,108 < 2% in two papers59,109). These papers further confirm that using this approach to measuring effectiveness, there is evidence of positive outcomes for people who stutter across a range of intervention approaches.

Although stuttering frequency or severity measures were the most frequent outcome data reported, a smaller number of papers considered wider effects on the person who stutters or self-rated perceptions of stuttering. One feedback and technology paper92 used PSI scores and found that the significant effect of the technology immediately post fitting was not maintained at follow-up. The ‘cognitive’ interventions group (as may be expected) tended to use a wider range of measures to evaluate efficacy. They indicated that the intervention could have an impact on not only stuttered speech but also self-perceptions and attitudes. De Veer et al. ,18 for example, reported large ESs on anxiety and locus of control. In the ‘speech motor and cognitive category’, Lawson et al. 75 found change in PSI scores with reduction in avoidance the greatest area of change.

Dose–response outcomes

We endeavoured to examine the included literature to explore whether or not the number of hours of intervention could be linked to outcomes for people who stutter. The heterogeneity in measures used and variation in time points assessed made this type of analysis problematic; however, in order to explore this potential relationship we tabulated papers that included statistical analysis (p-values or ESs; Table 16). It can be seen that not only did different measures preclude drawing robust conclusions regarding a relationship, but also there was a substantial body of literature which reported that intervention hours varied between individuals receiving the same intervention. Interventions varied from only a few hours (mostly technology and feedback) to > 75 hours. If interventions included residential components, time was estimated as being more than ‘working day hours’ as many reported including evening social activities. However, for these studies, potentially all waking hours could be considered intervention hours making the estimate of ‘more than 75 hours’ potentially considerably below that actually received.

Conclusions regarding any dose–response relationship are therefore limited owing to challenges extracting accurate information from studies and issues of differing outcome measurement. Interestingly, there was little discussion regarding how the contact hours had been determined for interventions with pre-designed schedules. Papers relating to the LP reported that individuals with more severe stuttering tended to require a greater number of contact hours and those with greater time since onset tended to require more sessions. There was no clear evidence that increasing contact hours for all participants led to more positive outcomes. Dose–response relationships seemed to be associated with characteristics of the people who stutter rather than the type and dosage of intervention.

Long-term effects

Fifty-one papers reported data at follow-up of 1 year or more following intervention. 16,20,21,25,27–29,37,38,40,41,45–47,49,51,56,57,59,61,63,66,69,71–74,77,79–82,84,90,91,96,97,100,103,104,106–110,112,114–117,122 The feedback and technology group, perhaps unsurprisingly owing to the nature of the interventions, tended to report immediate follow-up, with the effect of this technology demonstrated as a ‘quick fix’ method to reduce the percentage of syllables stuttered. Studies predominantly reported effects in laboratory rather than everyday settings; however, there was evidence to demonstrate its value in situations such as using the telephone. The other typologies provide evidence of long-term benefits (e.g. one study73 reports 71–86% of participants maintained gains) although there is evidence of fading of effect for many studies and substantial individual variation in the degree of preservation of effect. One study38 described that 46% of variance in effect at long-term follow-up between participants could not be accounted for.

Having considered the range of outcomes measured and examined evidence of positive outcomes across intervention approaches, a key question resulting from the review is that if these diverse types of intervention can all be effective then what is it about interventions that achieve change, what is the active ingredient that may be common across these differing programmes? Having analysed the intervention typologies and the outcomes, we then turned to the qualitative findings to seek further understanding of how these interventions may lead to their intended outcomes and whether or not the individual variation in outcomes reported may be explained by this literature.

Study design

All but one of the included studies used semistructured interviews to collect data. 128–151 Two studies used repeated interviews,139,140 one included telephone as well as face-to-face interviews139 and one study used only web conferencing technology to collect interview data. 116 One study supplemented interviews with questionnaires,128 two studies added focus groups131,136 and a further study used only focus groups. 152 Eleven studies described a phenomenological approach to analysis,116,129,134,135,138,139,142,148–151 and two used Grounded Theory. 146,147 Seven studies reported using Thematic Analysis,116,128,137,140,142,143,152 one Framework Analysis136 and one Content Analysis. 144 A further three did not report a specific method of analysis, although two of these described stages representing a thematic approach. 132,133,141

All included studies examined lived experiences and coping strategies of being a person who stutters or a spouse/mother/parent of a person who stutters. One study focused on reflections of childhood experiences of stuttering by adults who stutter. 141 One study focused on ethnicity and another on the client/therapist relationship. 137,143 Five studies assessed views following therapy or self-help conference,116,136,139,140,145 two of these explored parental views about the LP139,140 and two explored adult experiences of PS therapy. 134,135 One study reported adolescent perceptions of a range of therapies including an intensive week-long course in PS, an intensive week-long course in Smooth Speech, individual PS therapy and the Camperdown approach. 141 One study assessed the experience of late recovery from stuttering. 128

Population

Of the included papers, four were published in the UK,131,136,140,150 11 in the USA,116,128,137,138,142,146–149,151 three in Canada,130,132,133 six in Australia129,134,135,139,141,152 and two in South Africa. 133,143 Populations in all but one study were adults (one assessed the views of adolescents and young adults who stuttered). 141 Of these, two studies included parents (one included mothers only) of children who stutter139,140 and two assessed the views of spouses129,130 (one included both fluent and dysfluent partners129) of people who stutter. One study focused on the interaction between stuttering and ethnicity, with a sample of African American males. 137 No papers described participants as being clutterers.

Views about interventions aimed at children

Eleven of the included papers provided data about experiences and views following childhood interventions for developmental stuttering;116,128,129,131–135,138,141,142 however, none of these studies included participants that were children at the time of the research. Nine papers included retrospective data relating to childhood experiences in adult samples. 128,132,136–138,142–145 Three further papers described parental experiences of supporting their children through therapy. 139,140,149 Two papers139,140 evaluated parental experiences of implementing the LP, which was developed specifically for early stuttering intervention. The programme comprises parental training to give appropriate and timely feedback to the child on stuttering instances. The treatment mechanism is reinforcement of non-stuttered speech through parental praise, which needs to outweigh reminders not to stutter by at least five instances to one. 123

Views about interventions aimed at adolescents

Three included papers provided evidence relating to therapy for stuttering during adolescence. 128,132,141 Compared with the extent of available evidence about childhood and adult therapy, evidence about adolescent therapy was limited.

One participant in the Anderson and Felsenfield study128 attributed their recovery from stuttering as partly to therapy received during childhood, but also to taking public speaking courses during adolescence. Another interviewee132 reported starting to receive therapy in grade 8, although there was dissatisfaction that therapy focused on techniques without addressing psychological issues. Only one study reported on perceptions about therapy experiences in the adolescent age group. 141

Views about interventions aimed at adults

Nineteen of the included papers reported on studies that focused on the adult experience of stuttering and therapy. 116,128–137,143–148,150,151 Of these, three explored experiences of PS therapy,132–135 one116 evaluated a 15-day residential intensive programme and another focused on the implications of group therapy. 150 The remaining studies included some data about specific therapies and strategies although the study did not focus on any intervention in particular.

One study151 included attendees of a self-help conference and another study136 compared a group that had received therapy and one that had not. Plexico et al. 148 assessed characteristics of speech therapists from the attendee perspective, and Boberg and Boberg130 interviewed wives of stutterers. Two studies focused on ethnicity and its interaction with stuttering. 138,143

Interventions aimed at older adults

One included study152 focused on stuttering experiences in older adults, although no specific therapy was assessed. Many of the issues for older adults will be shared with adults in general, so this section only comments on the impact of older adulthood on stuttering intervention.

Stuttering across the lifespan

From the 26 included papers, there was evidence to suggest that people who stutter are impacted by life stages in relation to how they address their stuttering. 116,128–152 To some extent, the way that stuttering is addressed is influenced by interactions with other people in day-to-day situations. It is also influenced by growing maturity and acceptance of ones self as a stutterer.