Intervenciones no farmacológicas para la tartamudez en niños de hasta seis años
Resumen
Antecedentes
La tartamudez, o tartamudeo, afecta a la capacidad del niño de hablar con fluidez. Se trata de un trastorno de la comunicación muy frecuente, que afecta al 11% de los niños hasta los cuatro años de edad. La tartamudez puede caracterizarse por la repetición de sonidos, de parte de la palabra o de toda la palabra, por la prolongación de los sonidos o por el bloqueo de los sonidos o del flujo de aire. Los momentos de tartamudez también pueden ir acompañados de comportamientos no verbales, como una tensión visible en la cara del hablante, parpadeos o inclinaciones de cabeza. La tartamudez también puede afectar negativamente a la actividad conductual, social y emocional.
Objetivos
Objetivo principal
Evaluar los efectos inmediatos y a largo plazo de las intervenciones no farmacológicas para la tartamudez en los desenlaces del habla, las actitudes comunicativas, la calidad de vida y los posibles efectos adversos en niños de hasta seis años.
Objetivo secundario
Describir la relación entre los efectos de la intervención y las características preliminares de los participantes (es decir, la edad del niño, el coeficiente intelectual, la gravedad, el sexo y el tiempo transcurrido desde el inicio de la tartamudez).
Métodos de búsqueda
El 16 de septiembre de 2020 se hicieron búsquedas en CENTRAL, MEDLINE, Embase, PsycINFO, CINAHL, en otras nueve bases de datos y en dos registros de ensayos, y en Open Grey el 20 de octubre de 2020. No hubo restricciones de idioma, año de publicación ni tipo de publicación. También se buscó en las listas de referencias de los estudios incluidos y se solicitaron datos de ensayos no publicados a los autores de los estudios publicados. Se realizaron búsquedas manuales de resúmenes de congresos y programas de congresos relevantes.
Criterios de selección
Se incluyeron los ensayos controlados aleatorizados (ECA) y los ensayos controlados cuasialeatorizados que evaluaron intervenciones no farmacológicas para la tartamudez en niños de hasta seis años. Los comparadores elegibles fueron ninguna intervención, lista de espera o tratamiento habitual.
Obtención y análisis de los datos
Se utilizaron los procedimientos metodológicos estándares previstos por Cochrane.
Resultados principales
Se identificaron cuatro ECA elegibles, de los cuales todos compararon el Programa Lidcombe con un grupo control en lista de espera. En total, en los cuatro estudios incluidos participaron 151 niños de dos a seis años de edad. En el Programa Lidcombe, los padres y su hijo acuden a un logopeda en una clínica. Un estudio realizó las consultas clínicas por teléfono. En cada consulta clínica, se enseñó a los padres cómo llevar a cabo el tratamiento en casa. Dos estudios se realizaron en Australia, uno en Nueva Zelanda y uno en Alemania. Dos estudios duraron nueve meses, uno 16 semanas y uno 12 semanas. La frecuencia de las consultas clínicas y de las sesiones prácticas en casa varió dentro del programa. El Rotary Club de Wiesbaden, Alemania financió parcialmente un estudio; el National Health and Medical Research Council de Australia financió otro. Un estudio no informó sobre sus fuentes de financiación y otro comunicó que no recibió financiación alguna para el ensayo.
Los cuatro estudios informaron sobre el desenlace de la frecuencia de tartamudez. Un estudio también informó sobre la eficiencia del habla, definida como la tasa de articulación. Ningún estudio informó sobre los otros desenlaces predeterminados de esta revisión, es decir, la gravedad de la tartamudez; las actitudes comunicativas; los dominios emocionales, cognitivos o psicosociales; ni sobre los efectos adversos.
El Programa Lidcombe dio como resultado una frecuencia de tartamudez menor en porcentaje de sílabas tartamudeadas (% SS) que un grupo control en lista de espera tras la intervención, a las 12 semanas, a las 16 semanas y a los nueve meses desde la asignación al azar (diferencia de medias [DM] ‐2,16; intervalo de confianza [IC] del 95%: ‐3,48 a ‐0,84; cuatro estudios, 151 participantes; p = 0,001; evidencia de certeza muy baja).
Sin embargo, como el Programa Lidcombe está diseñado para durar de uno a dos años, ninguno de los participantes de estos estudios había terminado el programa de intervención completo en ninguno de los momentos de obtención de datos.
Se evaluó que la frecuencia de tartamudez tenía un alto riesgo de sesgo general debido al alto riesgo de sesgo en al menos un dominio en tres de los cuatro estudios incluidos, y que había cierta preocupación de sesgo general en el cuarto, debido a alguna preocupación en al menos un dominio.
Se encontró evidencia de certeza moderada de un estudio que muestra que el Programa Lidcombe podría aumentar la eficiencia del habla en los niños pequeños.
Solo un estudio informó sobre los desenlaces en el seguimiento a largo plazo. El efecto a largo plazo de la intervención no pudo resumirse, ya que faltaban los resultados de la mayoría de los niños del grupo control. Sin embargo, se realizó una comparación intragrupo entre la media de % SS en el momento de la asignación al azar y la media de % SS en el momento del seguimiento ampliado que mostró una reducción significativa de la frecuencia de tartamudez.
Conclusiones de los autores
Esta revisión sistemática indica que el Programa Lidcombe podría dar lugar a una menor frecuencia de tartamudez y una mayor eficiencia del habla que un grupo control en lista de espera en niños de hasta seis años de edad tras la intervención. Sin embargo, estos resultados deben interpretarse con cautela debido a la certeza muy baja y moderada de la evidencia, y al alto riesgo de sesgo identificado en los estudios incluidos. Por lo tanto, se necesitan más estudios de investigadores independientes, para evaluar los efectos inmediatos y a largo plazo de otras intervenciones no farmacológicas para la tartamudez en comparación con un grupo control sin intervención o en lista de espera.
PICO
Resumen en términos sencillos
Tratamientos sin medicamentos para la tartamudez en niños de hasta seis años
¿Cuál fue el objetivo de esta revisión?
El objetivo de esta revisión fue averiguar si los tratamientos sin medicamentos para la tartamudez pueden mejorar la fluidez del habla, las actitudes comunicativas del niño y el efecto en su calidad de vida, y los posibles efectos perjudiciales en niños de hasta seis años de edad, tanto a corto como a largo plazo. Se recopilaron y analizaron todos los estudios relevantes para responder a esta pregunta y se encontraron cuatro estudios.
Mensajes clave
El Programa Lidcombe podría dar lugar a una menor frecuencia de tartamudez y una mayor eficiencia del habla (es decir, número de palabras o sílabas habladas por minuto) en niños pequeños tras recibir la cantidad de tratamiento incluida en los estudios. Sigue sin conocerse el impacto del programa impartido en su totalidad, ya que ningún estudio comunicó los desenlaces de los niños que habían completado el programa, el cual está diseñado para durar de uno a dos años.
Solo un estudio informó sobre cómo funcionaba el tratamiento a largo plazo, pero el efecto del tratamiento no pudo resumirse, ya que faltaban los resultados de la mayoría de los niños del grupo control.
Se necesitan más estudios de alta calidad que evalúen los tratamientos de la tartamudez para niños pequeños, incluyendo estudios que informen sobre una gama más amplia de desenlaces y que evalúen otros tratamientos además del Programa Lidcombe.
¿Qué estudió la revisión?
La tartamudez, o el tartamudeo, como se le llama a veces, es un trastorno comunicativo frecuente que suele aparecer cuando los niños tienen entre dos y cuatro años de edad. La tartamudez puede caracterizarse por la repetición de sonidos individuales del habla, partes de palabras o palabras enteras; el alargamiento involuntario de los sonidos del habla; o los bloqueos del habla. Los momentos de tartamudez también pueden ir acompañados de tensión visible en la cara o la voz del hablante, guiños o inclinaciones de cabeza. La tartamudez puede tener un efecto negativo en cómo se sienten las personas consigo mismas y en cómo viven sus vidas. Por lo tanto, es importante identificar tratamientos eficaces para la tartamudez en los niños pequeños para reducir la posibilidad de que experimenten estos efectos negativos.
¿Cuáles son los principales resultados de la revisión?
Se encontraron cuatro estudios, de los cuales todos compararon a niños pequeños que recibieron el Programa Lidcombe con niños pequeños en lista de espera. En tres estudios, los niños de la lista de espera no recibieron ningún tratamiento hasta que se terminó el estudio. En el cuarto, los niños de la lista de espera podían recibir tratamiento de su logopeda local si lo deseaban. Al final del estudio, los padres de siete niños (35%) del grupo control informaron de que su hijo había recibido algún tratamiento, ya sea el Programa Lidcombe u otro programa de tratamiento llamado "Easy Does It", mientras estaba en la lista de espera.
Hubo 151 niños de dos a seis años de edad en los cuatro estudios. En el Programa Lidcombe, un logopeda realizaba consultas clínicas presenciales con el niño y los padres en una clínica o hablaba con ellos por teléfono. Durante estas consultas, se enseñó a los padres a llevar a cabo el tratamiento en casa en sesiones prácticas diarias de 10 a 15 minutos. Dos estudios se realizaron en Australia, uno en Nueva Zelanda y uno en Alemania. Dos estudios duraron nueve meses, uno 16 semanas y uno 12 semanas. El Rotary Club de Wiesbaden, Alemania financió parcialmente un estudio; el National Health and Medical Research Council de Australia financió otro. Un estudio no declaró fuentes de financiación y otro informó de que no recibió financiación alguna para el ensayo.
Los cuatro estudios notificaron los efectos del tratamiento sobre la frecuencia de tartamudez. Un estudio también notificó la eficiencia del habla. No se encontraron estudios que analizaran los efectos de los tratamientos para la tartamudez sobre la gravedad de la tartamudez; las actitudes comunicativas; el desarrollo emocional (cómo un niño reconoce, expresa y maneja sus sentimientos), cognitivo (cómo un niño piensa, explora y resuelve los problemas) o psicosocial (cómo las necesidades individuales de un niño se vinculan con las necesidades o demandas de la sociedad); ni los efectos secundarios.
Los estudios incluidos sugieren que el Programa Lidcombe podría reducir la frecuencia de la tartamudez en niños pequeños en comparación con un grupo control en lista de espera (evidencia de calidad muy baja). Un estudio también notificó que el Programa Lidcombe podría aumentar la eficiencia del habla en niños pequeños en comparación con un grupo control en lista de espera (evidencia de calidad moderada).
Solo un estudio hizo un seguimiento de los niños de hasta cinco años desde el inicio del tratamiento, y por aquel entonces, faltaban los resultados de la mayoría de los niños del grupo control. Por lo tanto, se desconoce si los beneficios del tratamiento duraron en el tiempo.
¿Cuál es el grado de actualización de esta revisión?
Se buscaron estudios que se hubieran publicado hasta el 16 de septiembre de 2020. Se hizo una búsqueda en otra base de datos el 20 de octubre de 2020.
¿Qué fiabilidad tiene la evidencia generada por esta revisión?
La calidad de los resultados relacionados con la frecuencia de tartamudez fue muy baja y la de los resultados relacionados con la eficiencia del habla fue moderada. Esto significa que no está claro que el efecto del tratamiento siga siendo el mismo si se añadiesen más estudios a la revisión. Se necesitan más estudios que comparen diferentes tratamientos para la tartamudez con un grupo control en lista de espera para conocer con mayor certeza el efecto del tratamiento para la tartamudez.
Authors' conclusions
Summary of findings
| The Lidcombe Program compared to wait‐list control group for stuttering in children aged six years and younger (see footnotes a and b) | ||||||
| Patient or population: children aged six years and younger who stutter Setting: home. Parents and child met with the speech and language therapist (SLT) once a week. Parents subsequently delivered the treatment in daily practice sessions with the child in their home Intervention: The Lidcombe Program Comparison: Wait‐list control group | ||||||
| Outcomes | Anticipated absolute effects* (95% CI) | Relative effect | № of participants | Certainty of the evidence | Comments | |
|---|---|---|---|---|---|---|
| Risk with Wait‐list control group | Risk with The Lidcombe Program | |||||
| Stuttering frequency (% SS) | The mean stuttering frequency was 1.9 to 6.3 % SS d | MD 2.16 % SS lower | ‐ | 151 | ⊕⊝⊝⊝ | All studies measured stuttering frequency by calculating percentage syllables stuttered (% SS), based on audio recordings in 3 studies, and audio and video recordings in 1 study, from 3 or 4 different speaking situations. A low score in percentage indicated improvement in stuttering. The evidence suggests that the Lidcombe Program may result in a lower stuttering frequency than a wait‐list control group. |
| Stuttering severity | not pooled | not pooled | not pooled | (0 studies) | ‐ | Not reported. |
| Speech efficiency (syllables per second) | The mean speech efficiency was 3.28 syllables per second | MD 0.3 syllables per second higher | ‐ | 46 | ⊕⊕⊕⊝ | 1 study measured speech efficiency by calculating the number of non‐stuttered syllables per second of speaking time, based on audio and video recordings from 2 different speaking situations. A higher score in percentage related to improvement. |
| Communication attitudes | not pooled | not pooled | not pooled | (0 studies) | ‐ | Not reported. |
| Impact of interventions for stuttering on emotional, cognitive and psychosocial domains, or quality of life | not pooled | not pooled | not pooled | (0 studies) | ‐ | Not reported. |
| Adverse effects | not pooled | not pooled | not pooled | (0 studies) | ‐ | Not reported. |
| *The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). | ||||||
| GRADE Working Group grades of evidence | ||||||
| See interactive version of this table: https://gdt.gradepro.org/presentations/#/isof/isof_question_revman_web_416961006131617056. | ||||||
| a. All studies used the Lidcombe Program to assess the non‐pharmacological stuttering interventions for stuttering. The Lidcombe Program is an intervention programme for stuttering in young children aimed to reduce stuttering. During the practice sessions at home, parents reinforced target speech behaviours to increase the child's fluency. | ||||||
Background
Description of the condition
Stuttering, or stammering as it is referred to in some countries, is a communication disorder that affects the fluency of speech. It can be characterized by verbal and non‐verbal behaviours. Verbal stuttering behaviours can include repetitions of individual sounds, parts of words, or whole words; or fixed postures, which may include prolongations whereby the production of an individual sound is extended, or blocks whereby the individual is temporarily unable to produce any sound (Ambrose 1999). Stuttering may also be accompanied by interjections of extra sounds or words such as 'uh' in order to terminate or avoid the stuttering moment (Guitar 2019). Non‐verbal stuttering behaviours may also occur during stuttering moments and may include visible facial or neck tension, eye blinks, head nods or audible inspiratory airflow (Guitar 2019; Packman 1998).
Developmental stuttering is common; however, cumulative incidence rates reported in the literature vary. While Månsson 2000 found that 5% of children stutter at three years of age, Dworzynski 2007 reported that the incidence of stuttering was 1.1% at two years of age, increasing to 2.0% at three years and 3.5% at four years. At seven years of age, the incidence decreased to 1.6%. Later, Reilly 2013 reported that 11.2% of children started stuttering by four years of age. Among people with intellectual disabilities, there are no incidence data, but the prevalence of stuttering is often described to be higher than that reported in the general population (Eggers 2018; Schieve 2009; Van Borsel 2007). Differences in inclusion criteria, definitions of stuttering and age of participants might explain the differences in incidence rates reported across studies (Yairi 2013). Stuttering typically begins suddenly between two and four years of age and coincides with the period of language development when children begin to combine words to form phrases (Reilly 2009). There are reports, however, of stuttering onset occurring as early as 12 months of age (Reilly 2009).
The exact cause of stuttering remains unknown (Chang 2020). There is consensus, however, that stuttering is a polygenic disorder and is underpinned by deficits in the neural processing of speech production and initiation of speech motor programming (Chang 2020; Kang 2010; Kraft 2012; Neumann 2003; Smith 2017). While evidence to describe the genetic component of the disorder is accumulating, to date, the specific genes involved have not been identified (Yairi 2013). There is also a consensus that stuttering moments can be triggered by factors within the child or their environment (Smith 2017). Packman 2012 proposes that features of spoken language increase the motoric demand on an individual's motor speech system (e.g. varied syllabic stress or increased syntactic complexity). These features can then trigger stuttering moments in people who have an unstable motor speech system due to an underlying neural processing deficit. Packman also described modulating factors that influence the threshold at which stuttering moments may be triggered. These are intrinsic emotional and cognitive capacities (e.g. feeling anxious in specific situations or executing cognitively demanding tasks such as recalling a story). Packman purports that these modulating factors may explain variation in an individual's stuttering across different situations as well as variation between individuals (Packman 2012).
While the majority of children recover from stuttering by school age, either spontaneously or facilitated by intervention, up to 35% of children who stutter at a young age develop persistent stuttering (Kefalianos 2017; Yairi 1999). Children who do recover typically do so at least 12 months after stuttering onset (Reilly 2013). While it is not currently possible to predict which children will recover from stuttering and which children will persist with stuttering, there is a consensus in the literature that boys are more likely to have persistent stuttering than girls (Yairi 2013). Across studies (see Sugathan 2020 for a review), a sharp decline in disfluencies one year after stuttering onset has been found to be associated with a higher chance of recovery.
Stuttering is a multifaceted disorder that can have an adverse effect on a person's quality of life (QoL) (Craig 2009; Koedoot 2011). Studies have consistently found that young children who stutter report more negative attitudes towards their own speech compared to their peers who do not stutter (Clark 2012; Vanryckeghem 2005; Węsierska 2015; for a review, see Guttormsen 2015). Investigations of behavioural, emotional and social development have yielded less‐consistent findings. Based on parent reports, Reilly 2013 found no behavioural, emotional or social functioning differences between children who do and do not stutter at four years of age, while McAllister 2016 and Briley 2019 reported that children who stutter have poorer behavioural, emotional and social functioning compared to their non‐stuttering peers at three (McAllister 2016) and five (Briley 2019) years of age. These inconsistencies may relate to different age ranges of participants included in the studies or the different instruments used to investigate participant's behavioural, emotional and social functioning.
For children who continue to stutter, more adverse effects of stuttering have been identified. Stuttering is associated with a heightened risk of anxiety disorders (Iverach 2016), bullying (Blood 2007), and reduced popularity among peers (Berchiatti 2020; Davis 2002). In the school‐aged population, Lankman 2015 revealed a moderate association between clinician‐rated stuttering severity and how the children perceived stuttering to affect their QoL. This result indicated that factors other than stuttering severity can impact a child's QoL. For example, Yaruss 2016 found an association between negative reactions to stuttering and adverse effects of stuttering on QoL. Several studies have found that children who stutter may experience teasing and social rejection as a consequence of stuttering (Ezrati‐Vinacour 2001; Langevin 2009; Langevin 2010). Subsequently, these negative experiences impede the child's QoL.
By adulthood, people who stutter and seek therapy have a 16‐fold increased odds of meeting the criteria for social anxiety disorder compared to non‐stuttering adults (Iverach 2009). They are also at increased risk for several other mental health disorders, including generalized anxiety disorder and depression (Iverach 2009). Stuttering can affect an individual's educational and occupational attainment (Klein 2004; O'Brian 2011). Considering the potential negative impact that stuttering can have on many aspects of a person's life, and the fact that it is currently not possible to predict who will recover from stuttering, it is important to determine effective management approaches during the early years for reducing the likelihood of children developing persistent stuttering.
Description of the intervention
Non‐pharmacological interventions for stuttering in children aged from birth to six years (hereafter referred to as young children) vary in terms of their aims, contents, dosages and theoretical underpinnings. These interventions are often conducted by or under guidance of speech and language therapists (SLTs), in preschools or speech and language therapy clinics. In many intervention programmes, the child's parents or carers are instructed by an SLT to implement the intervention at home, during structured practise sessions or spontaneously throughout the day. Most interventions do not specify a prescribed amount of treatment for each child. Instead, intervention continues until the aims of that specific programme have been achieved. Consequently, the length of the intervention is largely influenced by each child's responsiveness to the intervention.
While some intervention programmes for young children aim to reduce or eliminate stuttering, others aim to establish an acceptable level of stuttering as agreed by the parent and SLT (Sidavi 2010). Some intervention programmes also include an additional aim to enhance the child's QoL and develop positive communication attitudes (Baxter 2015).
Non‐pharmacological interventions for stuttering in young children have been categorized in different ways within the literature. A commonly used distinction is between direct and indirect interventions (De Sonneville‐Koedoot 2015; Nippold 2018; Sidavi 2010). Direct approaches involve teaching the child's parents to alter the child's speech. Programmes that are typically described as direct are those that include strategies such as reducing the child's linguistic output within structured conversations, reinforcing target speech behaviours or teaching the child speech restructuring strategies to control their fluency (Franken 2007; Onslow 2021;Trajkovski 2019; Waelkens 2018). Indirect approaches involve training the child's parents to create a fluency‐inducing environment. Programmes that are typically categorized within this approach include strategies such as teaching the parents to reduce their own speech rate or asking the child fewer questions (Franken 2007; Kelman 2017). This dichotomous categorization, however, is arguably problematic as intervention programmes often include a combination of direct and indirect approaches (Franken 2007; Kelman 2017; Waelkens 2018). Considering this, it may be more accurate to describe intervention programmes for stuttering as lying on a continuum, according to the proportion of direct and indirect strategies used.
How the intervention might work
Many non‐pharmacological interventions for stuttering have been developed for young children. However, research to date has not thoroughly investigated which components in available intervention programmes are responsible for instigating change in a child's stuttering (Bernstein Ratner 2018; Nippold 2018), or whether effect of intervention is reflected in neurological changes. Theoretical frameworks may, however, provide some insight into the mechanisms involved or how intervention programmes target causes of stuttering. Interventions typically referred to as direct treatment are designed to allow the child to practice the target motor speech pattern to elicit stutter‐free speech consistently during treatment time. Such interventions aim to modify features of spoken language that trigger the moments of stuttering (Packman 2012). The targeted motor speech pattern may involve changing features of spoken language such as syllabic stress (see Trajkovski 2006; Trajkovski 2019). Interventions typically described as indirect focus more on changing the child's environment to induce fluency, and are based on the theory that reducing the demands placed on the child's speech will increase their capacity to produce stutter‐free speech (Starkweather 1990). Thus, indirect treatments primarily focus on the modulating factors that determine the threshold for when moments of stuttering are triggered (Packman 2012). Packman 2012 theorises that reducing environmental stressors such as time pressure can reduce physiological arousal and raise the threshold for when stuttering moments are triggered.
Assessing the impact of intervention
Stuttering pre‐ and post‐test is most commonly measured by assessing speech outcomes, such as the frequency of stuttering moments or stuttering severity. Stuttering frequency can be measured by the SLT by calculating the percentage of syllables stuttered, percentage of words stuttered or number of stuttering‐like disfluencies (SLD) using a smartphone application or computer software program. SLD is a clinical measure of stuttering that includes descriptions of typical stuttering behaviours in young children. This measure states that all children produce SLDs and that a child is classified as stuttering if they produce more than three SLDs per 100 words (Yairi 2005). Stuttering severity can be rated on a Likert scale (Onslow 2021; Yairi 2015), based on the SLT and parents' observations of the child's speech during clinic sessions, or the parents' observations of their child throughout the day. Alternatively, or in addition, stuttering severity can be measured with a Stuttering Severity instrument – Fourth Edition (SSI‐4) calculating an overall severity score based on the frequency and duration of stuttering moments, and a description of the types of stuttering behaviours produced based on samples of the child's speech (Riley 2009). Speech efficiency can be measured to assess changes in speech, by measuring speech rate or articulation rate. Speech rate is typically measured by calculating the number of words or syllables spoken per minute. To ensure that a representative measure of a child's stuttering is obtained, measures of stuttering frequency and severity can be based on multiple 10‐ to 15‐minute speech samples, and preferably be video recorded (Guitar 2019; Riley 2009). Multiple speech samples across different speaking situations and with different communication partners account for variability of stuttering. Video recordings, as opposed to audio‐only recordings, facilitate the identification of any non‐verbal stuttering behaviours. To conduct a holistic assessment of the effect of intervention, parent‐report and child self‐report questionnaires can be used to assess the child's reactions to their own stuttering or communication and QoL (Boey 2009; Langevin 2010; Vanryckeghem 2006).
Adverse effects
The above mentioned measures can be used to assess both positive and adverse effects of intervention. In addition, qualitative descriptions of treatment experiences can be used to detect potential adverse effects of intervention. For example, Packman 2006 examined parents' experiences delivering the intervention, and found that some parents reported that the use of particular behavioural reinforcements during the intervention evoked a negative reaction, such as irritation or frustration, from their child.
Why it is important to do this review
As children grow older, stuttering becomes less tractable. This is thought to be attributable to declining neural plasticity of the brain with increasing age (Burke 2006). Interventions for young children who stutter therefore typically differ in aim and content from those of older children and adults (Ambrose 2006; Onslow 2019). Interventions for older children and adults typically focus on methods to reduce stuttering in specific situations or to communicate more effectively (Ambrose 2006), whereas interventions for young children typically focus on reducing or eliminating stuttering completely (Guitar 2019). This is thought to be attributable to age‐related changes in the structure and neural activity of the brain. Researchers have not established to date what neural differences are potential core deficits associated with stuttering and which differences are related to compensatory neural changes that occur in the brain after due to persisting stuttering behaviour (Burke 2006; Chang 2020). Interventions for young children who stutter therefore typically differ in aim and content from those of older children and adults (Ambrose 2006; Onslow 2019). Interventions for older children and adults typically focus on methods to reduce stuttering in specific situations or to communicate more effectively (Ambrose 2006), whereas interventions for young children typically focus on reducing or eliminating stuttering completely (Guitar 2019). Further, considering the high level of recovery among young children who stutter without intervention, synthesized results from randomized controlled trials (RCTs) comparing intervention to no intervention, wait list or management as usual, for young children specifically, will provide SLTs, stakeholders and parents with valuable information about the actual need for treatment (Kefalianos 2017).
In addition to the need to evaluate the evidence for the effects of interventions for stuttering in young children specifically, we note that previous systematic reviews in this area had several limitations; for a summary of existing systematic reviews and meta‐analyses see Appendix 1. Nye 2013 used a restricted selection of databases in their literature search that may have limited the number of studies identified. The Nye 2013 review described no systematic procedures for identifying grey literature; however, they searched the ProQuest Dissertations & Theses Full Text database. Similarly, grey literature was neither included in the search strategies nor the reviews by Baxter 2015, Bothe 2006, or McGill 2019. Given that null findings are less likely to be published (Borenstein 2009), the omission of grey literature from these reviews may have biased results in studies that aimed to assess the effect of interventions. Finally, most prior systematic reviews did not evaluate the evidence with the same criteria as a Cochrane Review (determining risk of bias using GRADE).
No prior meta‐analyses have calculated long‐term overall effect sizes of stuttering interventions. Nye 2013 were unable to report follow‐up effects of interventions due to the absence of usable data in the included studies. Based on this, Nye 2013 recommended that future treatment studies assess the long‐term effects of interventions.
The latest published review that included a meta‐analysis of studies of young children who stutter was based on searches conducted in July 2012 (Nye 2013). As the field of non‐pharmacological interventions for stuttering is growing (Baxter 2015), a new and more comprehensive systematic review and meta‐analysis is warranted. The review will influence speech and language therapy practice and clinical decision making.
Objectives
Primary objective
To assess the immediate and long‐term effects of non‐pharmacological interventions for stuttering on speech outcomes, communication attitudes, quality of life and potential adverse effects in children aged six years and younger.
Secondary objective
To describe the relationship between intervention effects and participant characteristics (i.e. child age, IQ, severity, sex and time since stuttering onset) at pretest.
Methods
Criteria for considering studies for this review
Types of studies
We included RCTs in which groups were established by random allocation, and quasi‐RCTs where the method of allocation was known but not considered strictly random (e.g. allocation based on birth date).
We excluded cross‐over trials.
Types of participants
We included studies with children aged up to six years at recruitment, as well as studies that involved children over six years of age, provided we were able to isolate the data for children aged up to six years. We included studies regardless of time since stuttering onset, language(s) spoken or language abilities of the participants.
We included relevant studies of participants with physical or intellectual disabilities (e.g. Down's syndrome) provided the stuttering was developmental in nature and not acquired.
Types of interventions
We included studies of all forms of non‐pharmacological interventions for stuttering in young children, including the use of electronic devices or smart phone applications, compared to no intervention, a wait‐list control or management as usual. Wait‐list control refers to participants in the comparison group not receiving treatment within the trial while waiting for the intervention. Management as usual refers to the continuum of care pathways a child who stutters may experience outside of a trial, from no treatment to the different treatment approaches, strategies and programmes that are available for children in this age group. We excluded studies comparing any given intervention with a named or known intervention from the review.
Types of outcome measures
We included studies assessing stuttering with standardized tests or speech samples conducted by an SLT, or child and parent reports completed at pre‐intervention (if available), post‐test, short‐term follow‐up (up to 12 months postintervention) and long‐term follow‐up (12 months or longer postintervention).
When studies reported on measures other than our primary and secondary outcomes, we described such measures in an additional table (Table 1), but did not report on them when synthesizing the results.
| Outcomes | Study | Participants reported | Time point measured | Description of outcome measure | Result |
|---|---|---|---|---|---|
| Acceptability/satisfaction of intervention
| Children in the intervention group only | Mean time since randomization 5 years (range 3.5–7.0 years) | In a custom‐made, standardized telephone interview, parents were asked how satisfied they were with the Lidcombe Program treatment. The possible answers were "Very satisfied", "Satisfied" or "Not satisfied". | Most parents were satisfied or very satisfied with the Lidcombe Program, but 1 was not satisfied. | |
| Children in the control group | In a custom‐made, standardized telephone interview, the parents of 2 children reported that their child had received the Lidcombe Program. The study report did not describe how the parents' satisfaction with Lidcombe program was assessed. | Both parents were very satisfied with the Lidcombe Program. | |||
| Children in the intervention group only
| On completion of Stage 1 of the Lidcombe Program
| A custom‐made questionnaire was sent to parents. In the questionnaire, parents were asked to respond to 9 statements about the acceptability or satisfaction of the telehealth intervention (clinical visits by telephone). The possible answers were "Totally agree," "Agree," "Neither agree nor disagree," "Disagree," or "Totally disagree". The questionnaire included statements like "I had enough training to feel confident in responding to my child's stutter‐free speech" and "I had enough training to feel confident in responding to my child's stuttering". | Most parents reported the telephone intervention process had been a positive experience. No other information was provided. | ||
| Satisfaction with the child's speech
| Children in the intervention group only
| Mean time since randomization was 5 years (range 3.5–7.0 years) | In a custom‐made, standardized telephone interview, parents were asked how satisfied they were with their child's level of speech fluency since completing the Lidcombe Program treatment. Parents were asked to respond to the question by answering "very satisfied", "satisfied" or "not satisfied". | Most parents were satisfied or very satisfied with their child's speech, but 3 were not. | |
| Children in the intervention group only
| 6 and 12 months after completion of Stage 1 of the Lidcombe Program | A custom‐made questionnaire was sent to parents. In the questionnaire, parents were asked "How satisfied have you been with (child's name)'s speech during the past month?". The possible answers were "very dissatisfied", "dissatisfied", "neither satisfied nor dissatisfied", "satisfied" or "very satisfied". | All responding parents rated they were very satisfied with their child's speech during the previous month at both time points. | ||
| Stuttering since completing the Lidcombe Program | Jones 2005
| Children in the intervention group only
| Mean time since randomization was 5 years (range 3.5–7.0 years)
| In a custom‐made, standardized telephone interview, parents were asked if their child had stuttered since the completion of the Lidcombe Program, and if so, how the fluency compared to that before the Lidcombe Program. | 8 (40%) children had stuttered at some time during the previous month while 12 children (60%) had not. 10 (50%) children had stuttered at some time since completing treatment and 10 (50%) had not stuttered at all. 9/10 parents reported their child's fluency was better, much better, a lot better or amazingly better than before the Lidcombe Program. 1 parent reported the child's fluency was about the same |
| Children in the control group completing the Lidcombe Program | In a custom‐made, standardized telephone interview, parents were asked if their children had stuttered since the completion of the Lidcombe Program. | 2 children had not stuttered since completing the programme. | |||
| Treatment time | Children in the intervention group only | On completion of Stage 1 of the Lidcombe Program
| Treatment time was measured in the following ways.
| The mean number of telephone consultations required to complete Stage 1 of the Lidcombe Program was 49 consultations (SD 26.8, range 27–98). The mean number of weeks required for completion of Stage 1 of the Lidcombe Program was 62.9 (SD 31.9, range 28–121). The mean frequency of consultation in days was 7.7 days (SD 0.9, range 6.4–9). The mean duration of consultation time was 33.1 minutes (SD 9.0, range 26–52). The mean total clinician time per telephone consultation reported was 77.3 minutes (SD 14.7, range 66–107).
|
SD: standard deviation.
Primary outcomes
-
Stuttering frequency, measured by calculating percentage of syllables stuttered, percentage of words stuttered or SLDs, etc.
-
Stuttering severity, measured using severity rating scales or stuttering severity instruments (e.g. SSI‐4; Riley 2009), etc.
-
Speech efficiency, reported as speech rate, measured by calculating the number of words or syllables spoken per minute, or articulation rate, measured by calculating the number of words or syllables spoken per minute, excluding non‐speaking time (e.g. pauses and non‐speech sounds).
Secondary outcomes
-
Communication attitudes, measured by parent report or self‐report (e.g. KiddyCAT (Communication Attitude Test for Preschool and Kindergarten Children Who Stutter); Vanryckeghem 2006).
-
Impact of interventions for stuttering on emotional, cognitive and psychosocial domains, or QoL, measured by parent/caregiver report (e.g. Child Behavior Checklist; Achenbach 2000).
-
Adverse effects, as reported by the study authors, including negative self‐evaluation or less participation in communication. It should be noted that all other mentioned outcome measures could also capture adverse effects of intervention.
Search methods for identification of studies
The searches were shared between the Cochrane Information Specialist for Cochrane Developmental, Psychosocial and Learning Problems (CDPLP) and the review team. The Information Specialist searched the databases listed under Electronic searches, apart from Open Grey, which was searched by the review team in collaboration with a librarian at the University of Oslo. We ran all searches in November 2019 and September 2020, except Open Grey, which was searched in February and October 2020. All search strategies are shown in Appendix 2. The search was not limited by language, year of publication or type of publication. In Open Grey, the search strategy was not limited to RCTs, as a preliminary search showed few available studies.
Electronic searches
We searched the following databases and trials registers.
-
Cochrane Central Register of Controlled Trials (CENTRAL; 2020, Issue 9) in the Cochrane Library. Searched 16 September 2020.
-
MEDLINE Ovid (1946 to September week 1 2020).
-
MEDLINE In‐Process & Other Non‐Indexed Citations Ovid (1946 to 15 September 2020).
-
MEDLINE E‐Pub Ahead of Print Ovid (15 September 2020).
-
Embase Ovid (1974 to 15 September 2020).
-
CINAHL Plus EBSCOhost (1937 to 16 September 2020).
-
APA PsycINFO Ovid (1806 to September week 1 2020).
-
ERIC EBSCOhost (1966 to 16 September 2020).
-
Epistemonikos (www.epistemonikos.org; searched 16 September 2020).
-
Cochrane Database of Systematic Reviews (CDSR; 2020, Issue 9) in the Cochrane Library. Searched 16 September 2020
-
Scopus Elsevier (searched 16 September 2020).
-
ProQuest Dissertations & Theses A&I (searched 16 September 2020).
-
speechBITE (speechbite.com; searched 16 September 2020).
-
ClinicalTrials.gov (clinicaltrials.gov; searched 16 September 2020).
-
World Health Organization International Clinical Trials Registry Platform (trialsearch.who.int/; searched 16 September 2020).
-
Open Grey (www.opengrey.eu; searched 20 October 2020).
Searching other resources
One review author (ÅS) searched for studies not identified by the electronic searches by reviewing the reference lists of included studies as well as other relevant studies the group identified, book chapters and systematic reviews. Another review author (HH) handsearched the Journal of Fluency Disorders on 23 October 2020 (from 1974 to 2020), a stuttering‐specific journal, which so far is not included in Cochrane's Handsearched Journals List (Cochrane 2019).
One review author (ÅS) handsearched conference proceedings and programmes from the stuttering‐specific conferences listed below for conference abstracts for all available years on 14 September 2020.
-
The World Congress of Fluency Disorders (2000 to 2018).
-
The Croatian Clinical Symposium of Stuttering (2016).
-
The Oxford Dysfluency Conference (2014 to 2017).
-
The International Conference on Stuttering (2012 to 2015).
In addition, ÅS contacted relevant researchers identified through the search, to identify other published or unpublished data from completed or ongoing studies for possible inclusion.
Data collection and analysis
The following sections report the methods that we used in the review. Readers are referred to our protocol, Sjøstrand 2019, and Table 2 for descriptions and explanations of unused methods.
| Methods described in the protocol, Sjøstrand 2019, but not used in review | Explanation | |
|---|---|---|
| Data collection and analysis | Selection of studies In cases where it was unclear to the review authors whether ≥ 2 publications were based on the same sample, we planned to contact the study authors for clarification. | This was not unclear in the included studies. |
| Measures of treatment effect
| Continuous outcome data For studies using different outcome measures to assess the same construct (e.g. severity rating scales with different scale descriptors), we planned to calculate the standardized mean difference (SMD) and present it with 95% confidence interval. | All included studies used similar outcome measures for the outcomes that could be pooled in a meta‐ analysis. |
| Dichotomous outcome data We aimed to express dichotomous data as risk ratios (RR) and present them with 95% confidence intervals. Following guidance from Higgins 2021 (Section 6.4.1.1), we planned to convert any results reported as odds ratios (OR) to RR before interpreting them. If the data were actually continuous but had been measured in a dichotomous manner, we planned to transform the OR to an SMD effect size (e.g. measures under or over a specified stuttering severity or frequency score). We planned to calculate the SMD using Hedges' g, interpreting effect sizes ranging from 0.0 to 0.30 as small, more than 0.30 to 0.80 as moderate and > 0.80 as large (Cohen 1988). | No studies with dichotomous outcome data were included in the review. | |
| Unit of analysis issues
| Cluster randomization We planned to use an intraclass correlation coefficient (ICC) to convert studies to their effective sample size before using the data in the meta‐analysis. If the information to calculate an ICC was not available from the included study's report, or the study authors did not analyze the data appropriately, the first review author (ÅS) planned to contact the study authors to retrieve the information needed. If we were unable to obtain this information from the study authors, we planned to gather external estimates from similar studies (Campbell 2000). | No studies with these designs were included in the review. |
| Studies with multiple treatment arms We planned to combine the intervention groups to form a single intervention group before comparing it with the control group, in order to avoid double counting the participants in the control group. | ||
| Dealing with missing data | We did not expect to have the data to be able to use replacement values for data deemed missing at random. Thus, we planned to conduct a sensitivity analysis to determine the potential impact of the missing data on the findings of our review. | 3 of the 4 included studies were rated at high overall risk of bias. The remaining study, therefore, could not be pooled in a separate analysis. |
| Assessment of reporting biases | We planned to create funnel plots to assess reporting bias if data from ≥ 10 studies were pooled. When interpreting these results, we would have considered possible explanations for funnel plot asymmetry, including true heterogeneity of effect with respect to sample size, bias of small trials and publication bias (Deeks 2021). | Only 4 studies were included in the review. |
| Data synthesis | We aimed to use follow‐up data to assess maintenance of treatment effect. We planned to pool data separately for short‐term follow‐up (up to 12 months postintervention) and long‐term follow‐up (≥ 12 months postintervention). | Insufficient data were included in the review. |
| Subgroup analysis and investigation of heterogeneity | We planned to analyze subgroups by conducting a moderator analysis, to test for intervention effects by subgroup interactions. Further, we planned to explore possible sources of clinical heterogeneity, by conducting the following subgroup analyses.
| No studies included children with intellectual disabilities. 2 studies reported sufficient data on participants' characteristics needed to form the other subgroups analyses. The data sample was too small to conduct meaningful analyses. |
| Sensitivity analysis | We planned to perform the following sensitivity analyses, to explore the influence of publication status and overall risk of bias on effect size.
| No unpublished trials were included in the meta‐ analysis. 3 of the 4 included studies were rated at high risk of bias overall. The remaining study, therefore, could not be pooled in a separate analysis. |
Selection of studies
We imported all records identified by our electronic search into EndNote (EndNote 2014), and subsequently DistillerSR (DistillerSR 2008), for screening. We identified and excluded duplicate records.
The screening process consisted of two stages conducted in DistillerSR (DistillerSR 2008). The first stage involved screening the title and abstract of each record against the selection criteria (see Criteria for considering studies for this review). Two review authors (ÅS, EK) independently screened the records ensuring all records were double screened. All records that satisfied the selection criteria progressed to the second stage of the screening process. At this stage, the full text of each remaining record was double screened against the selection criteria. The same review authors (ÅS, EK) independently conducted this second stage of screening. MK replaced EK in screening the few cases where the text was written in German. The review team were fluent in English, Norwegian, Danish, Swedish and German and therefore were able to review studies written in any of these languages. Six colleagues outside of the review team with proficiency in Czech, Dutch, Finnish, French, Russian and Spanish screened 12 records considered relevant based on their title and abstract. Inter‐rater reliability was strong (Kappa 0.81) (Cohen 1988). Review authors (ÅS, EK) resolved conflicts regarding the eligibility of studies at both stages of the screening process by discussion. If needed, they consulted with a third author (K‐ABN).
We describe every step of the selection process in a PRISMA flow chart (Liberati 2009). See Figure 1.
Study flow diagram
To ensure we did not treat publications from the same study as independent studies, we used the first publication as the primary reference. We also reported on the latest publication providing follow‐up data (Borenstein 2009).
Data extraction and management
Two review authors (ÅS, HH) independently extracted the required data from the included studies onto a data extraction form that was piloted prior to extraction. They resolved any differences in opinion by discussion with a third review author (K‐ABN).
Based on the checklist provided in Table 5.3.a in the Cochrane Handbook for Systematic Reviews of Interventions (Li 2021), the data extraction form included the following categories.
-
Information about data extracted from reports (title, unique identifier and date of publication, etc.).
-
Eligibility criteria (reason for inclusion).
-
Study design and method (RCT or quasi‐RCT).
-
Participant characteristics (e.g. sample size, age, sex, country of origin, language spoken, presence of comorbidities, stuttering frequency at recruitment, history of stuttering in the family, time since stuttering onset).
-
Intervention programme details (e.g. name of intervention programme; intervention provider; method of delivery; timing, frequency and duration of clinical visits; timing, frequency and duration of treatment provided at home).
-
Outcomes and outcome measures (e.g. any measures related to primary or secondary outcomes; see examples of measures under Types of outcome measures).
-
Results (e.g. number allocated to each intervention group, missing participants, summary data for each intervention group, estimate of effect with confidence intervals (CI), P value, subgroup analyses).
-
Miscellaneous details (e.g. key conclusion(s) from study authors, comments from study authors, required correspondence(s) with study authors to retrieve data, review authors' own comments on study, possible conflicts of interest (for study authors), stakeholders).
Once the review team agreed on the collected data, one review author (ÅS) entered the relevant data into RevMan Web 2020. Another review author (HH) cross‐checked all entered data for accuracy.
Assessment of risk of bias in included studies
We assessed the risk of bias of each included study using the Cochrane risk of bias tool 2 (RoB 2) (Sterne 2019) on 12 May 2020. The nature of the effect of interest of this review was the effect of assignment to the intervention (Sterne 2019). Two review authors (ÅS, HH) used the Excel tool to implement RoB 2 to independently assess the risk of bias for outcomes reported immediately post‐test in the included studies across the following five domains: bias arising from the randomization process; bias due to deviations from intended interventions; bias due to missing outcome data; bias in measurement of outcome and bias in selection of the reported outcome. The specific outcomes of interest were the primary and secondary outcomes listed in the Types of outcome measures section.
Based on the five domains, each included study was given a judgement of overall risk of bias. We assigned the rating 'low risk of bias' if we considered studies at low risk of bias for all five domains in relation to a study outcome. We assigned a rating of 'some concern of bias' if we judged studies to be at some concern of bias in at least one domain in relation to a study outcome. We assigned a rating of 'high risk of bias' if we judged studies were at a high risk of bias in at least one domain in relation to a study outcome, or if we judged studies to have some concerns across multiple domains in a way that substantially reduced our confidence in the result (Sterne 2019). The review authors (ÅS, HH) resolved conflicting ratings by discussion. If needed, they consulted a third review author (K‐ABN).
Conflict of interest
The RoB 2 tool does not directly assess the potential conflict of interest of study authors. Instead, information about funding and conflicts of interest of study authors was extracted from each study in the data extraction phase and is described in the Characteristics of included studies table. Two review authors (ÅS, HH) independently evaluated the level of concern about conflicts of interest as 'notable concern about conflict of interest' or 'no notable concern about conflict of interest'. They also provided a rationale for their evaluations, as recommended by Boutron 2021. There were no disagreements between the two review authors. These results are presented in Table 3. After collecting the data in the included studies we identified a need to perform an post hoc sensitivity analysis to investigate whether trials with conflicts of interest had different intervention effect estimates, or more variable effect estimates, compared to trials without an identified conflict of interest.
| Study | Judgement | Rational for judgement |
|---|---|---|
| Notable concern about conflict of interest. | Undeclared conflict of interest: the second (corresponding) author, third and fourth authors are developers of the Lidcombe Programa (Onslow 2001). | |
| Notable concern about conflict of interest. | Undeclared conflict of interest: the second (corresponding) author and third authors are developers of the Lidcombe Programa (Onslow 2001). | |
| No notable concern about conflict of interest. | The study report stated that the first author is a member of the Lidcombe Program Trainer Consortiumb. No conflicts of interest are known for the other 2 authors. | |
| Notable concern about conflict of interest. | Undeclared conflict of interest: the second and third (corresponding) authors are developers of the Lidcombe Programa (Onslow 2001). |
aThe financial benefits of developing The Lidcombe Program are unknown. However, we acknowledge that the relationship could create a conflict of interest as subsequent research funding may be impacted by favourable results in publications. This could potentially impact study design and risk of bias in study results.
bThe Lidcombe Program Trainers Consortium is a non‐profit group dedicated to providing professional continuing education to speech‐language pathologists about how to use the Lidcombe Program (Lidcombe Program 2021).
Measures of treatment effect
All data from the included studies were continuous. In instances where included studies used the same continuous outcome measure, as was the case for the primary outcome stuttering frequency, we estimated the intervention effect using the mean difference (MD) and presented it with the 95% CI.
Unit of analysis issues
We identified no cluster‐randomized trials or studies with more than two treatment arms for inclusion in the review.
Dealing with missing data
We contacted the corresponding author of Lewis 2008 to obtain missing outcome data (standard deviations (SD) for stuttering frequency at post‐test); they provided the requested information. Similarly, we contacted one of the authors of Lattermann 2008, in order to obtain the SD for speech efficiency; they also provided the requested data. We contacted the corresponding author of Harris 2002 for information about group sample sizes at randomization. This information was not obtained as it had been deleted from their records according to conditions stated in their ethical approval. For this one study, in the absence of data concerning group sizes, we made the reasonable assumption that children dropped out equally from the two groups.
Where it was not possible to obtain the missing data (as in the case of missing outcome data due to attrition), we noted the missing data on the data extraction form and in the risk of bias table related to each outcome. We discussed the extent to which the reported missing data were likely to influence the results of the study in the Summary of main results, as suggested by Deeks 2021.
Assessment of heterogeneity
A descriptive account of clinical heterogeneity, that is the distribution of participant characteristics (e.g. age, stuttering severity), characteristics of interventions used (e.g. intervention programme) and variability in outcomes reported (e.g. results from different speech outcome measures) across studies is provided in Included studies. We were unable to examine this variability statistically in subgroup analyses due to lack of data.
Methodological heterogeneity is described as variability in study design and risk of bias in Effects of interventions. We identified one study with a different type of control group to the three other studies (i.e. child could receive intervention for stuttering outside the trial while on the wait list) and one study with a different intervention delivery model (i.e. treatment sessions delivered by telephone rather than in person). Therefore, we conducted two post‐hoc sensitivity analyses, repeating the meta‐analysis for the outcome stuttering frequency with each of the aforementioned studies excluded. The findings are described in Effects of interventions, and discussed in Potential biases in the review process.
To assess statistical heterogeneity, we used the Chi² test, which provides evidence that variation in effect is caused by heterogeneity and not by chance. The Chi² test, however, might have low power in the meta‐analysis if a small number of studies are included or the included studies use small samples, as was the case in this review. Therefore, as suggested by Deeks 2021, we used a P value of 0.10 to determine statistical significance. We also reported Tau² as a measure of between‐study variability when reporting the results of the random‐effects model. Furthermore, we assessed the degree of heterogeneity across studies using the I² statistic. We interpreted the I² statistic in line with Section 10.10.2 of the Cochrane Handbook for Systematic Reviews of Interventions (Deeks 2021) as follows.
-
0% to 40%: might not be important.
-
30% to 60%: may represent moderate heterogeneity.
-
50% to 90%: may represent substantial heterogeneity.
-
75% to 100%: considerable heterogeneity.
Statistically significant unexplained heterogeneity affects the certainty assessment conducted with GRADE (see Quality of the evidence). Therefore, we interpreted the results with caution.
Assessment of reporting biases
As our meta‐analysis included fewer than 10 studies, we did not create and examine a funnel plot to assess reporting bias, as planned in our protocol (Sjøstrand 2019).
Data synthesis
We undertook a meta‐analysis of intervention effects at immediately post‐test using RevMan Web 2020. We synthesized the data using a random‐effects model with inverse‐variance weighting. We used the weighted mean of the intervention effects estimated in the individual studies to calculate the intervention effect estimate (Deeks 2021). In an inverse‐variance approach to meta‐analysis, larger studies with smaller standard errors are weighted more heavily than smaller studies with larger standard errors. Weighting done in this way decreases uncertainty of the pooled effect estimate (Deeks 2021). A random‐effects model accounts for variations between studies by increasing the standard errors and CI (Shadish 2002).
We were unable to assess the maintenance of intervention effect at follow‐up, as described in our protocol (Sjøstrand 2019). For the studies that reported on outcomes at follow‐up, we provided a narrative summary of the available data.
Where we identified only one study reporting an outcome, as was the case for the outcome speech efficiency, we provided a narrative summary of the available data.
Subgroup analysis and investigation of heterogeneity
Only two small studies provided data on participant characteristics (Lattermann 2008; Jones 2005). Subgroup analysis, as planned in our protocol (Sjøstrand 2019) was not conducted due to the lack of data.
Sensitivity analysis
We repeated the meta‐analysis based on a fixed‐effect model to detect differences between the random‐effects and fixed‐effect calculations (Deeks 2021). We could not conduct the two other preplanned sensitivity analyses (i.e. to repeat the analysis excluding unpublished trials and excluding studies rated at high risk of bias overall) as no unpublished trials were included in the meta‐analysis, and three of the four included studies were at high risk of bias overall. With only one study remaining, we could not repeat the analysis.
Summary of findings and assessment of the certainty of the evidence
We created a summary of findings table to present the findings for the Lidcombe Program versus a wait‐list control group, presenting data for all reported outcomes described in Types of outcome measures, at immediately post‐test. See summary of findings Table 1. The table was created using GRADEprofiler Guideline Development Tool (GRADEpro GDT) through RevMan Web. As planned in our protocol (Sjøstrand 2019), we reported the outcomes at the first available post‐test time point from each study. In the included studies, those time points were 12 weeks (Harris 2002), 16 weeks (Lattermann 2008), and nine months (Jones 2005; Lewis 2008) postrandomization.
Two review authors (ÅS, HH) independently assessed the certainty of the evidence for each outcome, and assigned ratings of high, moderate, low or very low certainty, according to the presence of the five criteria listed below and depicted in MECIR Box 14.2.a in the Cochrane Handbook for Systematic Reviews of Interventions (Schünemann 2021). Differences in opinions were resolved by discussion with a third review author (K‐ABN). We justified all decisions to downgrade the certainty of the evidence in the footnotes of the table.
-
Overall risk of bias (methodological quality).
-
Directness of evidence.
-
Heterogeneity.
-
Precision of results.
-
Risk of publication bias.
Results
Description of studies
See Characteristics of included studies, Characteristics of excluded studies, and Characteristics of studies awaiting classification tables.
Results of the search
We identified 7721 records from the electronic searches and an additional 542 records from searching other sources. Records from the electronic searches were imported into Endnote. Endnote identified 2813 duplicates. The remaining 4908 were imported into Distiller (DistillerSR 2008). DistillerSR identified and discarded another 349 duplicates, leaving 4559 records from the electronic searches and 542 records from other searches to be screened.
We excluded 4981 records at title and abstract screening, and assessed the full texts of the remaining 120 records against our selection criteria (Criteria for considering studies for this review). From these 120 records, 114 records were excluded due to various reasons and only four (from five study reports) met the inclusion criteria for this review. One study is awaiting classification (see Characteristics of studies awaiting classification table), as we did not succeed in reaching the study authors to obtain more details about the study in order to assess its eligibility. We identified no ongoing studies. See Figure 1.
Included studies
Four studies (from five study reports) met the eligibility criteria for this review (Harris 2002; Jones 2005; Lattermann 2008; Lewis 2008). See the Characteristics of included studies table for full details of each study.
Design
All four included studies were RCTs, with a parallel group design, comparing intervention effects of the Lidcombe Program to a wait‐list control group (Harris 2002; Jones 2005; Lattermann 2008; Lewis 2008).
Sample size
The total number of participants at randomization in the four studies was 151 (Harris 2002; Jones 2005; Lattermann 2008; Lewis 2008).
Setting
Two studies were set in Australia (Harris 2002; Lewis 2008), one in New Zealand (Jones 2005), and one in Germany (Lattermann 2008).
In three studies, the child and their parent visited the SLTs at private, university‐ or hospital‐based speech and language clinics (Harris 2002; Jones 2005; Lattermann 2008). During these visits, parents were taught to conduct treatment at home during 10‐ to 15‐minute daily 'practice sessions' (Onslow 2021, p 8). In the fourth study, clinic visits were conducted between the SLT and parent(s) by telephone (Lewis 2008). Similar to the other studies, parents were instructed to deliver treatment to their child at home in daily, 10‐ to 15‐minute 'practice sessions' (Onslow 2021, p 8).
Participants
Age
Participants were aged six years or younger at recruitment (Harris 2002; Jones 2005; Lattermann 2008; Lewis 2008). Participants ranged in age from two to six years.
Gender
All four studies included both girls and boys (Harris 2002; Jones 2005; Lattermann 2008; Lewis 2008). However, all studies recruited more boys. Across all studies, there were 28 (19%) girls and 117 (81%) boys at randomization (not including withdrawals in Harris 2002).
Time since stuttering onset
All participants had been stuttering for a minimum of six months at randomization (Harris 2002; Jones 2005; Lattermann 2008; Lewis 2008).
Stuttering frequency at recruitment
Two studies only included participants who presented with a minimum of 3% syllables stuttered (SS) at pretest (Harris 2002; Lattermann 2008). A third study specified a minimum of 2% SS in their inclusion criteria (Jones 2005). Lewis 2008 reported no selection criterion related to stuttering frequency. The studies reported the following mean % SS at pretest: 8.60% SS with intervention and 8.40% SS with control in Harris 2002; 6.4% SS with intervention and 6.80% SS with control in Jones 2005; 9.50% SS with intervention and 7.50% SS with control in Lattermann 2008; and 6.70% SS with intervention and 4.50% SS with control in Lewis 2008.
Family history of stuttering
Three studies detailed the percentage of participants who reported a positive family history of stuttering. Harris 2002 reported that 62% (18) of their participants had a positive family history of stuttering (not including withdrawals). Lattermann 2008 reported that 43% (20) of the participants had a known positive family history of stuttering and Lewis 2008 reported that 73% (16) of the participants had a positive family history of stuttering. Jones 2005 reported that 44% (24) of the participants had a family history of recovery from stuttering.
History of prior treatment
One study, Lewis 2008, excluded children who had previously or were currently receiving intervention for stuttering. Similarly, Jones 2005 excluded children who had received intervention for stuttering within the past 12 months. Harris 2002 excluded children who had previously completed the Lidcombe Program or received a professional consultation regarding the Lidcombe Program. Lattermann 2008 did not set any selection criteria related to history of prior intervention.
Attrition
All four studies reported attrition at post‐test assessment (Harris 2002; Jones 2005; Lattermann 2008; Lewis 2008). In Harris 2002, six children (21%) did not attend the post‐test assessment due to the following reasons: "death in the family" (Harris 2002, p 209), failure to comply with experimental procedures, treatment being sought closer to home and loss of contact with the family. The number of dropouts was not reported separately for the intervention and control groups. We contacted the corresponding author of the study but they could not provide this information as it had been deleted from their records according to conditions set in the study's ethical approval. In Jones 2005, two children in the intervention group withdrew from the study due to relocation or loss of contact with the families, and five in the control group (13% total) withdrew due to major illness in the case of one child, and loss of contact with the other families. In Lattermann 2008, one child (2%) in the control group did not attend the nine‐month postrandomization assessment due to the birth of a sibling. In Lewis 2008, one child in the intervention group was withdrawn from the study due to their parent failing to comply with study requirements. In the control group three children (18% total) were withdrawn due to loss of contact with families or non‐compliance with study requirements.
Profile of speech and language therapists
SLTs led treatment in all studies. In clinic visits between the parent, child and SLT, parents were taught how to conduct treatment in practice sessions at home. In two studies, the clinic visits were led by the first author, an SLT with extensive experience delivering the Lidcombe Program (Lattermann 2008; Lewis 2008). In the two other studies, SLTs also led clinic visits, however neither study reported the number of SLTs included in the trial or the training and level of experience of the treating SLTs (Harris 2002; Jones 2005).
Intervention
The non‐pharmacological intervention for stuttering used in all included studies was the Lidcombe Program. All four studies reported that the intervention was conducted in line with the Lidcombe Program treatment manual (Onslow 2021).
The Lidcombe Program is conducted in two stages. The duration of the intervention depends on the child's progress in each of these stages. "During Stage 1, the parent and child visit the SLT in the clinic once a week. Based on what they learn in the clinic, a parent delivers treatment as instructed by the SLT at home with their child, in daily 10–15 minutes "practice sessions"" (Onslow 2021, p 8). Parents give the following verbal contingencies through conversations during play and in everyday interactions with their child: acknowledgement or praise for periods of stutter‐free speech; acknowledgement of stuttering or a request for the child to correct a moment of stuttering; praise for correct self‐evaluation of stutter‐free speech; and praise for spontaneous self‐correction of stuttering. The child progresses to Stage 2 if stuttering frequency remains low over three executive weeks. Once Stage 2 commences, treatment is gradually withdrawn and the frequency of clinic visits gradually decreases over 12 months, provided stuttering frequency remains low.
In all included studies, the primary trials ended before all participants had completed both stages of the Lidcombe Program. Two studies were conducted for nine months: in Jones 2005 14 (52%) children in the intervention group reached Stage 2 within this timeframe, while seven (78%) children in Lewis 2008 reached Stage 2 in the same time frame. The third study was conducted for 16 weeks and three (13%) children in the intervention group progressed to Stage 2 (Lattermann 2008). All children in the intervention group continued with the intervention, as needed, at the completion of the study. Harris 2002 was conducted for 12 weeks. All participants were still in Stage 1 at the end of the trial and continued to receive the intervention, as needed, once the trial was finished. See Table 4 for an overview of the proportion of participants who completed Stage 1 in each study.
| Study | Post‐test time point | Percentage of children to complete Stage 1 (number) | Clinical visits needed to complete Stage 1 |
|---|---|---|---|
| 12 weeks postrandomization | None | All children were given 12 sessions. No information regarding whether the stage was completed at the post‐test time point. | |
| 9 months postrandomization | 52% (14) | No information | |
| 16 weeks postrandomization | 13% (3) | Mean 11 sessions; range 10–12 | |
| 9 months postrandomization | 78% (7) | Mean 49 sessions; range 26–98 |
aCriteria to complete Stage 1 in Lidcombe Program = less than 1% SS.
Comparator
The comparison group in all included studies was a wait‐list control group. In three studies, participants assigned to the control group did not receive any intervention during the trial (Harris 2002; Lattermann 2008; Lewis 2008). Once the trial ended however, these participants received the Lidcombe Program if they were still stuttering once they had reached the top of the wait‐list at the SLT clinic.
In the fourth study, parents of children assigned to the wait‐list control group were told at the start of the study that their child could receive treatment for stuttering during the trial at other SLT clinics, provided the treatment was not the Lidcombe Program (Jones 2005). At the end of the study, parents of seven children (35%) in the control group reported their child had received intervention outside the trial. Parents of four children reported that their child had received the Lidcombe Program while they were in the control group. Additionally, another parent reported that their child had received some components of the Lidcombe Program on an ad hoc basis, and the parents of two children reported that their children received the intervention 'Easy Does It' while assigned to the control group. Participants in the wait‐list control group received the Lidcombe Program at the end of the trial if they were still stuttering.
Outcomes
All four studies reported outcomes immediately after the trial had finished. While the Lidcombe Program is designed to end when stuttering is reduced to a minimal level (Onslow 2021), all studies had set a predetermined timeframe for intervention. As planned in our protocol (Sjøstrand 2019), we reported outcomes at the first available postintervention time point for each study. These time points were immediately post‐test in all studies, 12 weeks postrandomization for the participants in Harris 2002, 16 weeks postrandomization for the participants in Lattermann 2008, and nine months postrandomization for the participants in Jones 2005 and Lewis 2008.
Primary outcomes
Stuttering frequency
All studies reported stuttering frequency as their primary outcome measure. Stuttering frequency was measured by calculating the mean % SS based on several recordings of the children in different speaking situations. In two studies, the mean % SS was calculated based on audio recordings of the child in three different speaking situations taken outside the SLT clinic (Jones 2005; Lewis 2008). In a third study, the mean % SS was calculated based on three audio recordings taken outside the SLT clinic and one within the clinic (Harris 2002). Lattermann 2008 calculated one % SS based on two audio recordings from speaking situations outside the SLT clinic and one based on two video recordings within the SLT clinic. We calculated the mean of the two mean scores to enter into the meta‐analysis. Table 5 lists the different recorded speaking situations beyond and within the SLT clinic, used to calculate % SS in each study.
| Study | Number of recordings to base calculation of mean % SS | Description of speaking situation beyond the speech and language clinic | Description of speaking situation within the speech and language clinic |
|---|---|---|---|
| 4 audio recordings.
|
|
| |
| 3 audio recordings. |
| None. | |
| 4 recordings (audio or video recordings outside clinic, video recording at clinic). |
|
| |
| 3 audio recordings. |
| None. |
% SS: percentage syllables stuttered.
One study used "an electronic, button press counter and timer" for calculating % SS in each sample (Jones 2005, p 2). One study used a "custom‐made event counter" that automatically calculated % SS (Lewis 2008, p 142). The other two studies did not report the tool used to calculate % SS (Harris 2002; Lattermann 2008).
In three studies, an experienced SLT blinded to each participant's treatment status measured % SS (Harris 2002; Jones 2005; Lewis 2008). In the fourth study, the first author, who was aware of treatment status but was blinded to the time point the sample was recorded at, measured % SS (Lattermann 2008). In Lewis 2008, another independent and blinded SLT measured all recordings. In Harris 2002 and Lattermann 2008, an independent, blinded SLT measured 25% of the recordings. In Jones 2005, an independent blinded SLT measured 5% of the recordings. All studies reported a high degree of inter‐rater agreement for % SS (range of agreement r = 0.96 to 0.99). See the Characteristics of included studies table for more information.
Stuttering severity
No studies reported the effects of the intervention on stuttering severity at post‐test.
Speech efficiency
Lattermann 2008 reported the effects of the intervention on speech efficiency measured by articulation rate. Articulation rate was defined in the study as the mean number of non‐stuttered syllables per second of speaking time. Speaking time referred to the number of seconds elapsed to produce an utterance. Calculation of speaking time excluded non‐speech sounds, pauses exceeding 250 mseconds and segments with disfluency. The first author, who was aware of each participant's treatment status, measured this using a video recording taken at the clinic. Speech samples were "transcribed and coded orthographically" (Lattermann 2008, p 56) using the Computerized Language Analysis Program (CLAN). Data analysis is described in more detail in Characteristics of included studies table.
The remaining three studies did not report speech efficiency (Harris 2002; Jones 2005; Lewis 2008).
Secondary outcomes
Communication attitudes
None of the studies reported the effects of the intervention on communication attitudes.
Impact of interventions for stuttering on emotional, cognitive and psychosocial domains
None of the studies reported the impacts of the intervention on emotional, cognitive or psychosocial domains or QoL.
Adverse effects
None of the studies reported the effects of the intervention on adverse effects.
Other outcomes
No other outcomes were reported within any of the included studies comparing the effects of the intervention between the intervention group and control group.
Follow‐up
Lewis 2008 assessed stuttering frequency for children in the intervention group only, at 12 months, after completing Stage 1 of the Lidcombe Program. The mean number of weeks required for completion of Stage 1 was 62.9 (SD 31.9, range 28 to 121). Thus, the time point 12 months after completing Stage 1 varied across participants as it was dependent on the child's progress. The mean time that had lapsed at 12 months postcompletion of Stage 1 was two years three months postrandomization. The range was one year six months to three years four months postrandomization. The children in the control group who received the intervention after the wait‐list period were assessed for stuttering frequency at 18 months postrandomization. Stuttering frequency was measured as % SS based on recordings of the children in three different speaking situations.
Lewis 2008 also assessed the outcome of stuttering severity at six to 12 months after completion of Stage 1 of the Lidcombe Program for the intervention group only. Through a custom‐made questionnaire, the parents were asked to rate their child's stuttering severity during the past month, on a scale from 1 to 10 (1 represented "no stuttering", 2 represented "extreme mild stuttering" up to 10, which represented "extreme severe stuttering"). The other numbers were not described.
The second study report included from Jones 2005 measured stuttering frequency at extended follow‐up at a mean of five years (range 3.5 to 7.0 years) postrandomization. Stuttering frequency was reported as % SS and calculated based on an audio recording of a telephone conversation between each participant and an experienced SLT using a telephone recording jack. The SLT was not involved in the original trial but was not blinded to each participant's treatment status.
Jones 2005 also reported on the outcome of stuttering severity for the intervention group only at extended follow‐up. Through a custom‐made questionnaire, the parents were asked to rate their child's stuttering severity during the past month, on a scale from 1 to 10 (1 represented "no stuttering", 2 represented "extreme mild stuttering" up to 10, which represented "extreme severe stuttering"). The other numbers were not described.
Both Lewis 2008 and Jones 2005 reported other outcomes at different follow‐up time points. These outcomes are described in Table 1.
Funding sources
Harris 2002 did not report any funding sources and Jones 2005 reported that they did not receive any funding for the trial. Lattermann 2008 was partially funded by the Rotary Club, Wiesbaden, Germany; and Lewis 2008 was funded by the National Health and Medical Research Council of Australia.
Excluded studies
We excluded 114 full‐text reports that did not meet our Criteria for considering studies for this review: 12 were not about stuttering; 29 were not empirical studies; 32 included participants older than six years; nine included data for participants under six years that could not be isolated for analyses; 10 did not include a non‐pharmacological intervention; 19 were not RCTs; and three were RCTs comparing two stuttering interventions.
We reported the three RCTs comparing two stuttering interventions in the Characteristics of excluded studies table. We described the characteristics of eight other studies that were excluded at the abstract screening level, because they compared two or more interventions to each other in the Characteristics of excluded studies table. In accordance with recommendations made by Page 2021, we presented the characteristics of these 11 studies as they are "most likely to be considered eligible by readers" given that they are RCTs investigating stuttering interventions for young children.
Four studies compared altered versions of the Lidcombe Program to the original version of the Lidcombe Program (Arnott 2014; Bridgman 2016; Donaghy 2015; Koushik 2019).
Three studies compared The Lidcombe Program to a Demand and Capacities (DCM) approach (De Sonneville‐Koedoot 2015; Franken 2005; Sakata 2018).
One study conducted a three‐arm RCT comparing the Westmead Program, the Westmead Program with verbal contingencies and the standard Lidcombe Program (Trajkovski 2019).
One study compared Beilby Stuttering Therapy to Beilby Stuttering Therapy combined with The Curtin Early Childhood Stuttering Resilience Program (Druker 2019), and Druker 2020 compared a combination of Curtin University Stuttering Program (CUSP) and Evidence‐Based Parenting Support (EBPS) to CUSP for young children who stuttered and met the criteria for elevated attention deficit hyperactivity disorder (ADHD) symptoms. The two groups were also compared to a non‐randomized group of children who stuttered without ADHD symptoms who were also receiving CUSP.
One study compared a combination of acupuncture and speech and language therapy for stuttering in children to speech and language therapy for stuttering only (Figueras 2000).
Studies awaiting classification
We categorized one study as awaiting classification, see Characteristics of studies awaiting classification table (Paslawski 2018). Paslawski 2018 was a pilot study on animal assistance in fluency treatment. However, the record was identified in a conference programme, which included few details. We were unable to identify conference proceedings or a study report related to this study. The study seems to match the criteria of the present review in terms of types of studies, participants, and interventions. We sought clarification from the trial authors of the study but were unsuccessful in obtaining this information at the time of writing this review.
Risk of bias in included studies
Risk of bias judgements are summarized for the outcome stuttering frequency in Analysis 1.1 (Figure 2) and the outcome speech efficiency in Analysis 1.3 (Figure 3). The Excel tool used to implement RoB 2, and which is populated with detailed risk of bias assessment data, including consensus responses to the signalling questions, can be accessed at this link.
Forest plot of the comparison the Lidcombe Program versus wait‐list control group, outcome 1: stuttering frequency
Forest plot of the comparison the Lidcombe Program versus wait‐list control group, outcome 3: speech efficiency
We assessed all four studies at low risk of bias in the domain of 'Bias arising from the randomization process' (Harris 2002; Jones 2005; Lattermann 2008; Lewis 2008). All studies performed participant randomization independently.
There were two main differences between the risk of bias assessment for the outcomes stuttering frequency and speech efficiency. First, for stuttering frequency, two studies were at high risk of bias in the domain 'Bias due to missing outcome data' (Harris 2002; Lewis 2008). In contrast, for speech efficiency, Lattermann 2008 was at low risk of bias in the same domain. Second, for stuttering frequency, we assessed three studies at high overall risk of bias (Harris 2002; Jones 2005; Lewis 2008), due to high risk of bias identified in at least one domain in the case of Harris 2002 and Lewis 2008, and due to some concern identified in two domains in the case of Jones 2005. Contrastingly, we assessed the one study reporting on the outcome speech efficiency to be at some concern of overall bias due to some concern of bias in at least one domain (Lattermann 2008). A summary of the risk of bias assessment across outcomes for each study is provided within Effects of interventions.
Conflicts of interest
In all included studies, some of the authors had a direct connection to the Lidcombe Program, either as a Lidcombe Program Trainer Consortium member or as a program developer. Only Lattermann 2008 reported on this relationship. Table 3 summarizes our evaluation of potential conflicts of interest in the included studies. An post hoc sensitivity analysis excluding the three studies with notable concern about conflict of interest (Harris 2002; Jones 2005; Lewis 2008) showed that the MD in % SS between the groups at post‐test was higher in the study with no concern (−3.70, 95% CI −5.57 to −1.83; 46 participants) (Lattermann 2008), than in the meta‐analysis, suggesting that the conflict of interest did not affect the results.
Effects of interventions
See: summary of findings Table 1.
The Lidcombe Program versus a wait‐list control group
Primary outcomes
Stuttering frequency
All four studies reported the effect of the Lidcombe Program on stuttering frequency by calculating percentage syllables stuttered (% SS) (Harris 2002; Jones 2005; Lattermann 2008; Lewis 2008).
Results suggest that the Lidcombe Program may result in lower stuttering frequency than a wait‐list control group (P = 0.001). Based on a random‐effects model, the MD in % SS between the groups at post‐test was −2.16. (95% CI −3.48 to −0.84; 4 studies, 151 participants; P = 0.001; Analysis 1.1; Figure 2; summary of findings Table 1).
Heterogeneity
There was a substantial level of statistical heterogeneity (Tau² = 1.09; Chi² = 8.00, degrees of freedom (df) = 3 (P = 0.05); I² = 63%). We were unable to explore this heterogeneity in subgroup analyses, as stated in our protocol (Sjøstrand 2019), due to insufficient data being provided in the studies. All studies had similar study characteristics (type of intervention, outcome measures, participant age range and SLT profile; see Included studies). Lewis 2008, however, delivered the intervention in a different setting and participants had a lower level of stuttering frequency pre‐intervention. We found a change in the MD between groups when repeating the analysis with Lewis 2008 excluded, and a large change in statistical heterogeneity. The MD in % SS between the groups at post‐test was −2.78 (95% CI −3.81 to −1.75; 3 studies, 129 participants; P < 0.00001). Statistical heterogeneity might not be important (I² = 0%). Jones 2005 differed from the other studies in the way the control group was organized. While on the wait‐list, parents in the control group were able to seek treatment for their children outside of the trial (provided it was not the experimental intervention). Parents of seven children (35%) did so, and four (20%) of the children received the Lidcombe treatment outside of the trial whilst assigned to the control group. We found small changes in the MD between groups when repeating the analysis with Jones 2005 excluded. The MD in % SS between the groups at post‐test was −2.14 (95% CI −4.04 to −0.25; 3 studies, 97 participants; P = 0.03). Statistical heterogeneity was substantial (I² = 72%).
Risk of bias
The results of our risk of bias assessments for the outcome stuttering frequency at post‐test are shown in Analysis 1.1.
Bias due to deviations from intended interventions
We assessed all four studies at low risk of bias in 'Bias due to deviations from intended interventions', for the outcome stuttering frequency (Harris 2002; Jones 2005; Lattermann 2008; Lewis 2008). The SLTs, participants and parents were not blinded to treatment status. Blinding was not possible due to the nature of the intervention used in all studies. There were no signs, however, of deviations from the intended intervention programme.
Bias due to missing outcome data
We assessed two studies at high risk of bias in the domain 'Bias due to missing outcome data' (Harris 2002; Lewis 2008). Missing outcome data were related to attrition in both studies. Attrition rates were 21% in Harris 2002 and 18% in Lewis 2008 at post‐test. We assessed Jones 2005, with a 13% attrition rate, at some concern for bias in this domain. Lattermann 2008 was at low risk of bias on this domain as data from only one child (2%) were missing at post‐test.
Bias in measurement of the outcome
We assessed all four studies at low risk of bias in the domain 'Bias in measurement of the outcome'(Harris 2002; Jones 2005; Lattermann 2008; Lewis 2008).
Bias in selection of the reported result
For all four studies, we had some concerns for bias in the domain 'Bias in selection of the reported result' (Harris 2002; Jones 2005; Lattermann 2008; Lewis 2008). None of the studies provided details for a prespecified analysis plan. Therefore, we were unable to assess whether the stuttering frequency data were analyzed as intended, prior to unblinded data being made available for analysis.
Risk of overall bias
Following the recommendations of Sterne 2019, we assigned the judgement high risk of overall bias for this outcome to Harris 2002; Lewis 2008, due to the high risk of bias identified for the domain 'Bias due to missing outcome data' and to Jones 2005 due to some concerns for bias identified for the domains 'Bias due to missing outcome data' and 'Bias in selection of the reported result'. We assigned Lattermann 2008 a rating of some concerns for overall bias for this outcome, due to some concerns for bias identified in 'Bias in selection of the reported result'. Based on our risk of bias assessment, we concluded that the Lidcombe Program resulted in lower stuttering frequency than the stuttering frequency observed when there was no intervention.
Follow‐up
Lewis 2008 reported that the mean % SS was approximately 0.4 (only reported in a graph) for children in the intervention group 12 months after completing Stage 1 of the Lidcombe Program (time lapsed ranged from one year six months to three years four months post‐randomization). This finding was not compared to outcomes in the control group as children in the control group were offered the same intervention at the end of the wait‐list period. A within‐group comparison was not conducted between nine months post‐randomization and at follow‐up for the children in the intervention group.
Further, the mean % SS for seven of 10 children in the control group who received treatment after a wait‐list was 0.66% SS (SD 0.53, range 0 to 1.5) at 18 months postrandomization. A within‐group comparison between the mean % SS (1.9) at nine months postrandomization and at 18 months postrandomization showed a statistically significant reduction in stuttering for the children in the wait‐list control group (paired t test, P < 0.01).
The second article from Jones 2005 reported that the mean % SS was 1.53 (range 0 to 14.8) for children in the intervention group at extended follow‐up. Mean time since randomization was five years (range 3.5 to 7.0 years). A within‐group comparison was not conducted between the mean % SS at nine months postrandomization and at follow‐up for the children in the intervention group. However, a within‐group comparison between the mean % SS at randomization and the mean % SS at the time of extended follow‐up showed a significant reduction in frequency of stuttering (paired t‐test: MD 55.5% SS; P < 0.01). A comparison could not be made between the intervention and control groups at follow‐up, as the children in the control group were offered the same intervention at the end of the wait‐list period and only eight children in the control group were contactable. Of those eight children, five had received no treatment since the trial ended. Among those, stuttering frequency ranged from 0.0% SS to 0.5% SS. A further two participants had completed the Lidcombe Program successfully and showed 0% SS. One participant was receiving the Lidcombe Program and had a frequency of stuttering of 12.3% SS.
Stuttering severity
No studies reported the effects of the Lidcombe Program on stuttering severity at post‐test.
Follow‐up
Lewis 2008 reported that parents of children in the intervention group rated their child's stuttering severity as 1 "no stuttering" for the previous month at 12 months after completion of Stage 1 of the Lidcombe Program (range one year six months to three years four months postrandomization). These data were not provided for the control group.
The second article from Jones 2005 reported that parents of 60% of the children in the intervention group rated their child's stuttering severity to be 1 ("no stuttering" for the previous month), while 40% rated their child's stuttering severity to be between 2 and 6 for the previous month. These data were collected at a mean of five years (range 3.5 to 7.0 years) postrandomization.
Speech efficiency
One study reported the effect of the Lidcombe Program on speech efficiency (Lattermann 2008).
Articulation rate increased both in the intervention group from pretest to post‐test. The MD in non‐stuttered syllables per second between the groups at post‐test was 0.30 (95% CI −0.16 to 0.76; 1 study, 46 participants; Analysis 1.3; Figure 3; summary of findings Table 1).
Risk of bias
The results of our risk of bias assessments for the outcome of speech efficiency at post‐test are shown in Analysis 1.3.
Our risk of bias assessment indicated that Lattermann 2008 had a low risk of bias in four RoB 2 domains ('Bias arising from the randomization process', 'Bias due to deviations from intended interventions', 'Bias due to missing outcome' and 'Bias in measurement of the outcome').
Bias in selection of the reported result
We assessed Lattermann 2008 as having some concern for bias in the domain 'Bias in selection of the reported result'. The study did not provide a prespecified analysis plan, and therefore we were unable to assess whether speech efficiency data were analyzed as intended, prior to unblinded data being made available for analysis.
Risk of overall bias
Following the recommendations of Sterne 2019, we assigned a rating of some concerns for overall bias for this outcome in Lattermann 2008, due to our analysis that there were some concerns for bias in the domain 'Bias in selection of the reported result'. Based on our risk of bias assessment, we concluded that the Lidcombe Program may improve speech efficiency to a greater extent than the speech efficiency observed in a wait‐list control group.
Secondary outcomes
Communication attitudes
None of the included studies reported on the effects of the Lidcombe Program on communication attitudes.
Impact of interventions for stuttering on emotional, cognitive and psychosocial domains, or quality of life
None of the included studies reported on the effects of the Lidcombe Program on emotional, cognitive or psychosocial domains, or QoL.
Adverse effects
None of the included studies provided any information about adverse effects.
Other outcomes
As stated in our protocol (Sjøstrand 2019), we described outcomes other than the ones mentioned in Types of outcome measures, but did not reported them when synthesizing the results. No other outcomes were reported by any of the included studies comparing the effects of the intervention between the intervention group and control group at post‐test. However, both Jones 2005 and Lewis 2008 also reported other outcomes at different follow‐up time points. These outcomes are described in Table 1.
Relationship between intervention effects and participant characteristics
The relationship between intervention effects and participant characteristics was not investigated due to a lack of data.
Sensitivity analysis
Repeating the meta‐analysis using a fixed‐effect model
When we repeated the analysis for the outcome stuttering frequency using a fixed‐effect model, the MD was −1.84 (95% CI −2.60 to −1.09; 4 studies, 151 participants). Thus, there were only very small differences between results depending on whether random‐effects or fixed‐effect modelling was used in the meta‐analyses.
Repeating the analysis excluding studies where the overall risk of bias was high
Only one of the four included studies, Lattermann 2008, was not rated at high risk of bias overall for the primary outcome of stuttering frequency, and therefore the data could not be pooled in a separate analysis.
Discussion
Summary of main results
The objective of this review was to assess the immediate and long‐term effects of non‐pharmacological interventions for stuttering in young children.
We identified four studies for inclusion. All included studies investigated the Lidcombe Program and provided data for one of our primary outcome measures, stuttering frequency.
Our meta‐analysis found that the Lidcombe Program resulted in a lower stuttering frequency than a wait‐list control group immediately post‐test, and 12 weeks, 16 weeks, and nine months post‐randomization. None of the participants had finished the complete intervention program at any of these timepoints, as the Lidcombe Program is designed to take one to two years to complete (Onslow 2021).
Lattermann 2008 also investigated the Lidcombe Program's effect on speech efficiency. None of the studies reported on stuttering severity. Similarly, no studies reported on any of our secondary outcomes: communication attitudes; emotional, cognitive or psychosocial domains, or QoL; or adverse effects.
We did not find sufficient evidence to assess the maintenance of intervention effects long term. Jones 2005 reported intervention results at an extended follow‐up mean of five years (range 3.5 to 7.0 years) postrandomization, showing that less than half of the children had some signs of stuttering in the last month, and that stuttering frequency among participants was low on a group level. A meaningful comparison with the control group was not possible, as the study authors had lost contact with some participants, and others had received the intervention since the study finished. Therefore, there is insufficient evidence to ascertain whether the Lidcombe Program makes any difference to long‐term reductions in stuttering frequency compared to no intervention.
Overall completeness and applicability of evidence
We identified some methodological details within the included studies of this review that present possible challenges for generalizing the findings to the broader population of young children who stutter.
Applicability of the findings to locations and ethnicities other than those in the included studies is unclear. Two studies were conducted in Australia (Harris 2002; Lewis 2008), one study in New Zealand (Jones 2005), and one study in Germany (Lattermann 2008). In addition to the need for broader knowledge about effects across different local settings, exploration of the applicability of the Lidcombe Program and other stuttering interventions for other cultures must be considered. For example, behavioural reinforcement underpins the approach used in the Lidcombe Program. It is plausible that this approach may not be a typical form of interaction within some cultures between parents and their children. This may present some barriers to this intervention being adopted, either at all or in its current form, in some countries.
Due to the limited number of outcomes reported in the included studies, we were unable to draw conclusions about the effects of the intervention on any outcomes other than stuttering frequency, and partly speech efficiency. None of the included studies assessed the effects of the intervention on stuttering severity; communication attitudes; or emotional, cognitive or psychosocial domains, or QoL. Similarly, as none of the included studies reported on adverse effects, we were unable to draw any conclusions about the adverse effects of intervention in our review, other than a potential increase in stuttering frequency. Given that stuttering is a complex disorder, affecting a person's verbal communication and potentially the emotional, cognitive and psychosocial dimensions of life (Briley 2019; McAllister 2016), it seems critical that studies include these outcomes in their evaluation of interventions to obtain a more holistic understanding of an intervention's effects.
The only studies that satisfied eligibility criteria for inclusion in our review investigated the Lidcombe Program. These studies all involved people who developed the Lidcombe Program or who were members of the Lidcombe Program Trainers Consortium. As shown in the Characteristics of excluded studies tables, there are many other interventions for stuttering available to treat young children. We were unable to assess the effects of these other interventions as they failed to satisfy the selection criteria for our review.
The lengths of the included trials in this review were 12 weeks, 16 weeks and nine months. Approximately 18% of participants in the included studies completed Stage 1 of the Lidcombe Program (see Table 4). None of the participants had sufficient time within the trials to complete Stage 2 (Onslow 2021). Given this, we were unable to assess the effects of the Lidcombe Program when the programme had been completed. Continuing to evaluate the effects of the Lidcombe Program once the entire programme had been delivered to participants may have yielded even lower stuttering frequency for the intervention group, compared to the control group, reinforcing the effect of the Lidcombe Program further. Alternatively, over time, natural recovery may have reduced the difference in stuttering frequency between the two groups, given that approximately two‐thirds of children recover from stuttering by school entry, either naturally or facilitated by intervention (Kefalianos 2017).
Conclusions about the effects of the intervention can be made based on data obtained immediately post‐test. There was insufficient evidence available to determine the long‐term effects of the intervention. Limited evidence however, indicated that a notable number of children start to stutter again after receiving intervention (Jones 2005). In the absence of long‐term follow‐up data from clinical trials it is impossible to determine definitively whether any intervention is more effective than natural recovery of stuttering.
Quality of the evidence
Using the GRADEprofiler Guideline Development tool (GRADEpro GDT), we assessed the overall certainty of the evidence as very low for the outcome of stuttering frequency and moderate for the outcome of speech efficiency (see summary of findings Table 1).
The outcome of stuttering frequency was downgraded due to high risk of bias (missing outcome data), inconsistency due to moderate to substantial statistical heterogeneity (I² = 63%), and imprecision due to small sample sizes (fewer than 55 children in each study).
The outcome of speech efficiency was downgraded to moderate certainty in the domain of imprecision, primarily due to small sample size (less than 55 children in the included study). We were unable to downgrade for inconsistency for this outcome, as more than one study is needed to assess inconsistency in results.
As further research is conducted, and data from more studies can be added to these analyses, it is very likely this will have an important impact on our confidence in the estimated effects of the intervention.
The risk of publication bias was unclear, as we were unable to draw a funnel plot, due to the small number of studies (fewer than 10) included in the review (Deeks 2021).
Potential biases in the review process
We aimed to limit bias in the review process by using a broad search strategy, including numerous databases, conducting searches for grey literature and not imposing any language restrictions for potentially eligible studies. In order to minimize additional bias, the review authors (ÅS, EK) independently assessed studies for eligibility. Two review authors (ÅS, HH) independently extracted the data, assessed the risk of bias using the latest version of Cochrane's 'Risk of bias' tool, RoB 2 (Sterne 2019), and evaluated the concern about conflict of interest in the included studies (Boutron 2021).
We acknowledge that the decision to exclude RCTs that compared one intervention to another intervention may be considered as a selection bias to some readers. Our search indicated that 11 studies published since 2000 have used this design (see the Characteristics of excluded studies table). However, studies comparing two different interventions in an RCT do not address basic questions of effectiveness and answer a different research question to that stated in our review. Our review aimed to assess the immediate and long‐term effects of stuttering intervention compared to a wait‐list control group, no intervention or management as usual.
However, we acknowledge that including studies comparing the intervention to management as usual can restrict the present study in truly investigating the effect of the intervention compared to no intervention. In the present version of the review, we included no studies comparing intervention to management as usual. However, one of the included studies differed from the other studies in the way the control group was organized (Jones 2005). While on the wait‐list, parents in the control group were able to seek treatment for their children outside of the trial (provided it was not the experimental intervention). We acknowledge that the inclusion of this study arguably is a limitation to the review. Therefore, we conducted a sensitivity analysis repeating the meta‐analysis for the outcome stuttering frequency with Jones 2005 excluded, as described in Assessment of heterogeneity. The sensitivity analysis thus compared intervention to no intervention immediately post‐test. This comparison is essential to assess the effect of a stuttering intervention, particularly considering the high level of natural recovery among young children who stutter (Kefalianos 2017; Reilly 2013). The sensitivity analysis found only a minor difference in MD in % SS between an analysis comparing intervention to no intervention, and an analysis comparing intervention to a wait‐list in which some children had received some various treatment outside the trial.
Agreements and disagreements with other studies or reviews
To the best of our knowledge, there are no other systematic reviews of RCTs specifically focused on the effects of interventions for stuttering in young children. An overview of systematic reviews and meta‐analyses of non‐pharmacological interventions for stuttering in people of all ages has been included in Appendix 1. In addition to the mentioned reviews, we are aware that Onslow 2012 calculated an overall effect size based on the four studies included in our review (Harris 2002; Jones 2005; Lattermann 2008; Lewis 2008). This analysis, however, was published in a book chapter and was not based on a systematic search and review.
Among the studies presented in Appendix 1, we particularly recognize that there are similarities between our review and Nye 2013. Although the scope of their review was different (intervention for children under 18 years of age) and their search methods were more restricted (only one of the eight databases that were searched included grey literature), all four studies included in this review were also included in Nye 2013. Three of the four studies included in our review (Harris 2002; Jones 2005; Lattermann 2008) were synthesized in Nye 2013 and analyzed with the findings from a study that included older children. Because stuttering is less tractable in older children and adults, and the interventions for young children differ in aim and content from those of older children and adults (Ambrose 2006; Onslow 2019), we argue that it is crucial to separate these participant groups in analyses of intervention effects.
Our review also differs from that of Nye 2013, as we included a thorough risk of bias assessment for the outcomes reported in the included studies. Nye 2013 assessed the included studies using the Downs and Black Checklist (1998), a less comprehensive assessment than that used in Cochrane Reviews.
In addition, Nye 2013 was based on searches conducted up to July 2012. Therefore, we saw a need for a more recent analysis of the evidence for stuttering interventions for young children as we expected more studies to have been conducted. However, our comprehensive systematic search did not identify any new eligible studies. While we identified 11 RCTs of stuttering intervention (as listed in the Characteristics of excluded studies tables), none of these compared intervention to no intervention, a wait‐list control or management as usual. With a more recent and comprehensive search than that conducted in Nye 2013, we are able to conclude that only four studies have been published that are able to answer the question about the immediate effect of interventions for stuttering in young children. We discuss this finding in implications for research.
Study flow diagram
Forest plot of the comparison the Lidcombe Program versus wait‐list control group, outcome 1: stuttering frequency
Forest plot of the comparison the Lidcombe Program versus wait‐list control group, outcome 3: speech efficiency
Comparison 1: The Lidcome Program versus wait‐list control group for children who stutter aged between birth and six years, Outcome 1: Stuttering frequency
Comparison 1: The Lidcome Program versus wait‐list control group for children who stutter aged between birth and six years, Outcome 2: Stuttering severity
Comparison 1: The Lidcome Program versus wait‐list control group for children who stutter aged between birth and six years, Outcome 3: Speech efficiency
Comparison 1: The Lidcome Program versus wait‐list control group for children who stutter aged between birth and six years, Outcome 4: Communication attitudes
Comparison 1: The Lidcome Program versus wait‐list control group for children who stutter aged between birth and six years, Outcome 5: Impact of interventions for stuttering on emotional, cognitive and psychosocial domains, or quality of life
Comparison 1: The Lidcome Program versus wait‐list control group for children who stutter aged between birth and six years, Outcome 6: Adverse effects
| The Lidcombe Program compared to wait‐list control group for stuttering in children aged six years and younger (see footnotes a and b) | ||||||
| Patient or population: children aged six years and younger who stutter Setting: home. Parents and child met with the speech and language therapist (SLT) once a week. Parents subsequently delivered the treatment in daily practice sessions with the child in their home Intervention: The Lidcombe Program Comparison: Wait‐list control group | ||||||
| Outcomes | Anticipated absolute effects* (95% CI) | Relative effect | № of participants | Certainty of the evidence | Comments | |
|---|---|---|---|---|---|---|
| Risk with Wait‐list control group | Risk with The Lidcombe Program | |||||
| Stuttering frequency (% SS) | The mean stuttering frequency was 1.9 to 6.3 % SS d | MD 2.16 % SS lower | ‐ | 151 | ⊕⊝⊝⊝ | All studies measured stuttering frequency by calculating percentage syllables stuttered (% SS), based on audio recordings in 3 studies, and audio and video recordings in 1 study, from 3 or 4 different speaking situations. A low score in percentage indicated improvement in stuttering. The evidence suggests that the Lidcombe Program may result in a lower stuttering frequency than a wait‐list control group. |
| Stuttering severity | not pooled | not pooled | not pooled | (0 studies) | ‐ | Not reported. |
| Speech efficiency (syllables per second) | The mean speech efficiency was 3.28 syllables per second | MD 0.3 syllables per second higher | ‐ | 46 | ⊕⊕⊕⊝ | 1 study measured speech efficiency by calculating the number of non‐stuttered syllables per second of speaking time, based on audio and video recordings from 2 different speaking situations. A higher score in percentage related to improvement. |
| Communication attitudes | not pooled | not pooled | not pooled | (0 studies) | ‐ | Not reported. |
| Impact of interventions for stuttering on emotional, cognitive and psychosocial domains, or quality of life | not pooled | not pooled | not pooled | (0 studies) | ‐ | Not reported. |
| Adverse effects | not pooled | not pooled | not pooled | (0 studies) | ‐ | Not reported. |
| *The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). | ||||||
| GRADE Working Group grades of evidence | ||||||
| See interactive version of this table: https://gdt.gradepro.org/presentations/#/isof/isof_question_revman_web_416961006131617056. | ||||||
| a. All studies used the Lidcombe Program to assess the non‐pharmacological stuttering interventions for stuttering. The Lidcombe Program is an intervention programme for stuttering in young children aimed to reduce stuttering. During the practice sessions at home, parents reinforced target speech behaviours to increase the child's fluency. | ||||||
| Outcomes | Study | Participants reported | Time point measured | Description of outcome measure | Result |
|---|---|---|---|---|---|
| Acceptability/satisfaction of intervention
| Children in the intervention group only | Mean time since randomization 5 years (range 3.5–7.0 years) | In a custom‐made, standardized telephone interview, parents were asked how satisfied they were with the Lidcombe Program treatment. The possible answers were "Very satisfied", "Satisfied" or "Not satisfied". | Most parents were satisfied or very satisfied with the Lidcombe Program, but 1 was not satisfied. | |
| Children in the control group | In a custom‐made, standardized telephone interview, the parents of 2 children reported that their child had received the Lidcombe Program. The study report did not describe how the parents' satisfaction with Lidcombe program was assessed. | Both parents were very satisfied with the Lidcombe Program. | |||
| Children in the intervention group only
| On completion of Stage 1 of the Lidcombe Program
| A custom‐made questionnaire was sent to parents. In the questionnaire, parents were asked to respond to 9 statements about the acceptability or satisfaction of the telehealth intervention (clinical visits by telephone). The possible answers were "Totally agree," "Agree," "Neither agree nor disagree," "Disagree," or "Totally disagree". The questionnaire included statements like "I had enough training to feel confident in responding to my child's stutter‐free speech" and "I had enough training to feel confident in responding to my child's stuttering". | Most parents reported the telephone intervention process had been a positive experience. No other information was provided. | ||
| Satisfaction with the child's speech
| Children in the intervention group only
| Mean time since randomization was 5 years (range 3.5–7.0 years) | In a custom‐made, standardized telephone interview, parents were asked how satisfied they were with their child's level of speech fluency since completing the Lidcombe Program treatment. Parents were asked to respond to the question by answering "very satisfied", "satisfied" or "not satisfied". | Most parents were satisfied or very satisfied with their child's speech, but 3 were not. | |
| Children in the intervention group only
| 6 and 12 months after completion of Stage 1 of the Lidcombe Program | A custom‐made questionnaire was sent to parents. In the questionnaire, parents were asked "How satisfied have you been with (child's name)'s speech during the past month?". The possible answers were "very dissatisfied", "dissatisfied", "neither satisfied nor dissatisfied", "satisfied" or "very satisfied". | All responding parents rated they were very satisfied with their child's speech during the previous month at both time points. | ||
| Stuttering since completing the Lidcombe Program | Jones 2005
| Children in the intervention group only
| Mean time since randomization was 5 years (range 3.5–7.0 years)
| In a custom‐made, standardized telephone interview, parents were asked if their child had stuttered since the completion of the Lidcombe Program, and if so, how the fluency compared to that before the Lidcombe Program. | 8 (40%) children had stuttered at some time during the previous month while 12 children (60%) had not. 10 (50%) children had stuttered at some time since completing treatment and 10 (50%) had not stuttered at all. 9/10 parents reported their child's fluency was better, much better, a lot better or amazingly better than before the Lidcombe Program. 1 parent reported the child's fluency was about the same |
| Children in the control group completing the Lidcombe Program | In a custom‐made, standardized telephone interview, parents were asked if their children had stuttered since the completion of the Lidcombe Program. | 2 children had not stuttered since completing the programme. | |||
| Treatment time | Children in the intervention group only | On completion of Stage 1 of the Lidcombe Program
| Treatment time was measured in the following ways.
| The mean number of telephone consultations required to complete Stage 1 of the Lidcombe Program was 49 consultations (SD 26.8, range 27–98). The mean number of weeks required for completion of Stage 1 of the Lidcombe Program was 62.9 (SD 31.9, range 28–121). The mean frequency of consultation in days was 7.7 days (SD 0.9, range 6.4–9). The mean duration of consultation time was 33.1 minutes (SD 9.0, range 26–52). The mean total clinician time per telephone consultation reported was 77.3 minutes (SD 14.7, range 66–107).
| |
| SD: standard deviation. | |||||
| Methods described in the protocol, Sjøstrand 2019, but not used in review | Explanation | |
|---|---|---|
| Data collection and analysis | Selection of studies In cases where it was unclear to the review authors whether ≥ 2 publications were based on the same sample, we planned to contact the study authors for clarification. | This was not unclear in the included studies. |
| Measures of treatment effect
| Continuous outcome data For studies using different outcome measures to assess the same construct (e.g. severity rating scales with different scale descriptors), we planned to calculate the standardized mean difference (SMD) and present it with 95% confidence interval. | All included studies used similar outcome measures for the outcomes that could be pooled in a meta‐ analysis. |
| Dichotomous outcome data We aimed to express dichotomous data as risk ratios (RR) and present them with 95% confidence intervals. Following guidance from Higgins 2021 (Section 6.4.1.1), we planned to convert any results reported as odds ratios (OR) to RR before interpreting them. If the data were actually continuous but had been measured in a dichotomous manner, we planned to transform the OR to an SMD effect size (e.g. measures under or over a specified stuttering severity or frequency score). We planned to calculate the SMD using Hedges' g, interpreting effect sizes ranging from 0.0 to 0.30 as small, more than 0.30 to 0.80 as moderate and > 0.80 as large (Cohen 1988). | No studies with dichotomous outcome data were included in the review. | |
| Unit of analysis issues
| Cluster randomization We planned to use an intraclass correlation coefficient (ICC) to convert studies to their effective sample size before using the data in the meta‐analysis. If the information to calculate an ICC was not available from the included study's report, or the study authors did not analyze the data appropriately, the first review author (ÅS) planned to contact the study authors to retrieve the information needed. If we were unable to obtain this information from the study authors, we planned to gather external estimates from similar studies (Campbell 2000). | No studies with these designs were included in the review. |
| Studies with multiple treatment arms We planned to combine the intervention groups to form a single intervention group before comparing it with the control group, in order to avoid double counting the participants in the control group. | ||
| Dealing with missing data | We did not expect to have the data to be able to use replacement values for data deemed missing at random. Thus, we planned to conduct a sensitivity analysis to determine the potential impact of the missing data on the findings of our review. | 3 of the 4 included studies were rated at high overall risk of bias. The remaining study, therefore, could not be pooled in a separate analysis. |
| Assessment of reporting biases | We planned to create funnel plots to assess reporting bias if data from ≥ 10 studies were pooled. When interpreting these results, we would have considered possible explanations for funnel plot asymmetry, including true heterogeneity of effect with respect to sample size, bias of small trials and publication bias (Deeks 2021). | Only 4 studies were included in the review. |
| Data synthesis | We aimed to use follow‐up data to assess maintenance of treatment effect. We planned to pool data separately for short‐term follow‐up (up to 12 months postintervention) and long‐term follow‐up (≥ 12 months postintervention). | Insufficient data were included in the review. |
| Subgroup analysis and investigation of heterogeneity | We planned to analyze subgroups by conducting a moderator analysis, to test for intervention effects by subgroup interactions. Further, we planned to explore possible sources of clinical heterogeneity, by conducting the following subgroup analyses.
| No studies included children with intellectual disabilities. 2 studies reported sufficient data on participants' characteristics needed to form the other subgroups analyses. The data sample was too small to conduct meaningful analyses. |
| Sensitivity analysis | We planned to perform the following sensitivity analyses, to explore the influence of publication status and overall risk of bias on effect size.
| No unpublished trials were included in the meta‐ analysis. 3 of the 4 included studies were rated at high risk of bias overall. The remaining study, therefore, could not be pooled in a separate analysis. |
| Study | Judgement | Rational for judgement |
|---|---|---|
| Notable concern about conflict of interest. | Undeclared conflict of interest: the second (corresponding) author, third and fourth authors are developers of the Lidcombe Programa (Onslow 2001). | |
| Notable concern about conflict of interest. | Undeclared conflict of interest: the second (corresponding) author and third authors are developers of the Lidcombe Programa (Onslow 2001). | |
| No notable concern about conflict of interest. | The study report stated that the first author is a member of the Lidcombe Program Trainer Consortiumb. No conflicts of interest are known for the other 2 authors. | |
| Notable concern about conflict of interest. | Undeclared conflict of interest: the second and third (corresponding) authors are developers of the Lidcombe Programa (Onslow 2001). | |
| aThe financial benefits of developing The Lidcombe Program are unknown. However, we acknowledge that the relationship could create a conflict of interest as subsequent research funding may be impacted by favourable results in publications. This could potentially impact study design and risk of bias in study results. | ||
| Study | Post‐test time point | Percentage of children to complete Stage 1 (number) | Clinical visits needed to complete Stage 1 |
|---|---|---|---|
| 12 weeks postrandomization | None | All children were given 12 sessions. No information regarding whether the stage was completed at the post‐test time point. | |
| 9 months postrandomization | 52% (14) | No information | |
| 16 weeks postrandomization | 13% (3) | Mean 11 sessions; range 10–12 | |
| 9 months postrandomization | 78% (7) | Mean 49 sessions; range 26–98 | |
| aCriteria to complete Stage 1 in Lidcombe Program = less than 1% SS. | |||
| Study | Number of recordings to base calculation of mean % SS | Description of speaking situation beyond the speech and language clinic | Description of speaking situation within the speech and language clinic |
|---|---|---|---|
| 4 audio recordings.
|
|
| |
| 3 audio recordings. |
| None. | |
| 4 recordings (audio or video recordings outside clinic, video recording at clinic). |
|
| |
| 3 audio recordings. |
| None. | |
| % SS: percentage syllables stuttered. | |||
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1.1 Stuttering frequency Show forest plot | 4 | 151 | Mean Difference (IV, Random, 95% CI) | ‐2.16 [‐3.48, ‐0.84] |
| 1.2 Stuttering severity Show forest plot | 0 | 0 | Odds Ratio (IV, Random, 95% CI) | Not estimable |
| 1.3 Speech efficiency Show forest plot | 1 | 46 | Mean Difference (IV, Random, 95% CI) | 0.30 [‐0.16, 0.76] |
| 1.4 Communication attitudes Show forest plot | 0 | 0 | Peto Odds Ratio (Peto, Fixed, 95% CI) | Not estimable |
| 1.5 Impact of interventions for stuttering on emotional, cognitive and psychosocial domains, or quality of life Show forest plot | 0 | 0 | Peto Odds Ratio (Exp[(O‐E) / V], Fixed, 95% CI) | Not estimable |
| 1.6 Adverse effects Show forest plot | 0 | 0 | Odds Ratio (IV, Fixed, 95% CI) | Not estimable |
| Bias | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Study | Randomisation process | Deviations from intended interventions | Missing outcome data | Measurement of the outcome | Selection of the reported results | Overall | ||||||
| Harris 2002 |  | |  | |  | | ||||||
| Randomization was performed independently at the Clinical Trials Center of the National Health and Medical Research Council of Australia using a block design. Children allocated to the treatment group began intervention immediately after randomization. There was no baseline difference between the groups in % syllables stuttered scores. | ||||||||||||
| Neither children, parents or SLTs could be blinded to intervention assigned due to the nature of intervention. No reported deviations from intended intervention reported other than attrition. We see no reason to believe that the attrition is an direct consequence of the experimental context.
| ||||||||||||
| Outcome data were available for 79 % of the participants (23 out of 29 children) immediately post‐test. Three of the four reported reasons for missing outcome data show that missingness in the outcome could depend on its true value: failure to comply with experimental procedures, treatment sought closer to home and loss of contact with the family. The difference in attrition between the groups was not reported. | ||||||||||||
| An appropraite meassure for stuttering frequency was used (percentage syllables stuttered (% SS). % SS was calculated by an experienced SLT blinded to participant's treatment status. | ||||||||||||
| No information included about pre‐specified analysis plan. | ||||||||||||
| Following the recommendations of Sterne 2019 "overall high risk of bias" is assigned studies with high risk of bias in at least one domain. | ||||||||||||
| Jones 2005 | | | | | | | ||||||
| An independent central telephone randomisation service provided by the National Health and Medical Research Council (NHMRC) Clinical Trials Centre at the University of Sydney. Baseline characteristics of all participants show no large differences. | ||||||||||||
| Neither children, parents or SLTs could be blinded to intervention assigned due to the nature of intervention. | ||||||||||||
| Outcome data were available for 87% of the participants (47 of 54 children). In addition, two had missing data at nine months; data from 6 and 7 months were used instead. Reported reasons for missing outcome data did not provide evidence that missingness in the outcome depends on its true value. The reasons for not completing the trial included major illness in the case of one child, and families not being contactable mainly because of relocation | ||||||||||||
| An appropriate measure for stuttering frequency was used (percentage syllables stuttered (% SS). Observers blinded to treatment allocation assessed outcomes. | ||||||||||||
| No information reported about a pre‐specified analysis plan. | ||||||||||||
| Following the recommendations of Sterne 2019 "overall high risk of bias" is assigned studies with some concerns for multiple domains in a way that substantially lowers confidence in the result. | ||||||||||||
| Lattermann 2008 | | | | | | | ||||||
| The participating children were randomly assigned to treatment or wait‐contrast group by serial assignment according to the registration list. The children of the treatment group started intervention immediately. The participants were not matched on age, time since stuttering onset, gender, family history of stuttering including history of recovery, and language/phonological development, due to small sample. | ||||||||||||
| Neither children, parents or SLTs could be blinded to intervention assigned due to the nature of intervention. Deviation from intended intervention (one child withdrew) was probably not related to experimental context. | ||||||||||||
| Outcome data were available for 98% of the participants (45 of 46 children). | ||||||||||||
| An appropriate measure for stuttering frequency was used (percentage syllables stuttered (% SS). % SS was measured by the first author who was aware of treatment status but was blinded to the time point the sample was recorded at. 25% of the recordings were rated by an independent, blinded SLT with high level of agreement. | ||||||||||||
| No information reported about a pre‐specified analysis plan. | ||||||||||||
| Following the recommendations of Sterne 2019 "some concerns for overall bias" is assigned studies with some concerns for risk of bias in at least one domain. | ||||||||||||
| Lewis 2008 | | | | | | | ||||||
| Randomization was performed independently, with accepted methods, at the National Health and Medical Research Children in the control group were younger and presented with less severe stuttering at pre‐intervention. | ||||||||||||
| Neither children, parents or SLTs could be blinded to intervention assigned due to the nature of intervention. Reported to follow intention‐to‐treat principle. | ||||||||||||
| Outcome data were available for 82% of the participants (18 of 22 children). Reported reasons for missing outcome data showed that missingness in the outcome could depend on its true value: In the intervention group, one child was withdrawn because the parent failed to comply with study requirements. In the control group, one child was withdrawn because the parent could not be recontacted. Another 2 children withdrew from the control group as they believed their children were no longer stuttering and declined the request to provide a tape recording. There were also some differences between intervention groups in the proportions of missing outcome data: 1 participant had missing data in the intervention group compared to 3 participants in the control group. | ||||||||||||
| An appropriate measure for stuttering frequency was used (percentage syllables stuttered (% SS). All recordings creating the bases for the calculation of %SS were calculated by an independent and blinded SLT. | ||||||||||||
| No information reported about a pre‐specified analysis plan. | ||||||||||||
| Following the recommendations of Sterne 2019 "overall high risk of bias" is assigned studies with high risk of bias in at least one domain. | ||||||||||||
| Bias | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Study | Randomisation process | Deviations from intended interventions | Missing outcome data | Measurement of the outcome | Selection of the reported results | Overall | ||||||
| Lattermann 2008 | | | | | | | ||||||
| The participating children were randomly assigned to treatment or wait‐contrast group by serial assignment according to the registration list. The children of the treatment group started intervention immediately. The participants were not matched on age, time since stuttering onset, gender, family history of stuttering including history of recovery, and language/phonological development, due to small sample. | ||||||||||||
| Neither children, parents or SLTs could be blinded to intervention assigned due to the nature of intervention. Deviation from intended intervention (one child withdrew) was probably not related to experimental context. | ||||||||||||
| Outcome data were available for 98% of the participants (45 of 46 children). | ||||||||||||
| Speech efficency was defined as articulation rate and was measured by the first author aware of treatment status, based on a video recording taken at the clinic. Additionally, an independent person calculated the articulation rates from 18 randomly selected speech samples (20% of the data) to quantify inter‐observer reliability for these measurements. The correlation between both observers was r = .98. | ||||||||||||
| No information provided about a pre‐specified analysis plan. | ||||||||||||
| As suggested by Sterne 2019 we have assigned a rating of 'some concern of bias' to studies we judge to have some concerns of bias in at least one domain for the result in question. | ||||||||||||
