Speech-generating devices for children with complex communication needs: a systematic review and meta-analysis.

Abstract

Background

Speech-generating devices (SGDs) are a form of augmentative and alternative communication (AAC) that aim to support communication in individuals with complex communication needs. There have been few investigations of the efficacy of this approach.

Aim

This systematic review explored the current empirical landscape and aimed to determine the effects of interventions using high-tech, grid-based SGD for children with complex communication needs. We determine the efficacy of SGDs on communication outcomes and identified common methods, designs and outcome measures used to inform development of future trials in this space.

Method

A systematic search found 1648 studies. We included all study designs that involved children with any neurodevelopmental or genetic disorders, aged 0-18 years, who had complex communication needs (e.g. less than 50 spoken words). We excluded participants with acquired conditions or those with isolated speech disorders in the absence of any other neurodevelopmental condition. Interventions must have included the use of a high-tech, grid-based SGD, like Proloquo2Go on the iPad. Communication outcomes were extracted, including speech, language and non-verbal communication outcomes. Qualitative analysis was completed for both group and single-case experimental designs (SCEDs). A pooled analysis was completed for single cases that used comparable outcomes. The risk of bias was assessed and certainty of evidence derived from the GRADE approach.

Results

A total of 32 studies were included with 253 participants meeting criteria (age range 2-13 years). Most studies were SCEDs (28/32) conducted in the United States (22/32). The majority of studies included autistic participants (240/253). Other common groups included those with developmental disabilities (7/253), Down syndrome (6/253). The most common hardware was iPads (27/32). Proloquo2Go was the most common software (15/32).

There were 4 group studies (3 randomised trials, 1 quasi-experimental longitudinal study); 2 had high- and 2 moderate risk-of-bias. The studies were heterogeneous in trial design and intervention methods, e.g., use of different devices, software programs, implementation strategies (like least-to-most prompting, communication partner training) and outcome measures. There was positive change in communication outcomes across the 4 studies with varying effect sizes.

There were 28 SCEDs of varied designs and communication outcomes. These studies had low internal validity due to inadequate blinding of patients, therapists or assessors. Few studies measured maintenance, follow-up or generalisation of outcomes. Most (13/28) single-case studies measured independent and/or accurate requesting skills, while only 4 measured overall communication without focusing on function (e.g. request, respond, commenting), spontaneity or accuracy. The variability of outcomes meant only 16 studies were included in a quantitative pooled analysis, with the remaining described qualitatively. The studies that were pooled measured spontaneous and accurate use of communication devices and encompassed 53 single cases, which were extracted and used to compare baseline and intervention results. The results showed significant statistical difference (p < 0.001) in the positive direction (i.e. improvement) between the baseline and intervention phases. This change was consistent across different communication modes and functions.

Discussion

This systematic review identified significant heterogeneity in study designs for interventions using the SGDs. High-powered empirical evidence was limited, with only four group trials met criteria; none of which have been replicated. The variability in outcomes highlighted the need for increased standardisation of measures to consistently report the effects of the intervention. Majority of the SCED studies focused on measuring requesting (13/28) or responding (9/28) skills, while few studies holistically measured all or varied forms of communication (4/28). Future research should aim to measure progress in a range of communication functions.

There is also a dearth of AAC research in non-autistic populations, such as rare genetic disorders, even though communication disorders commonly have a genetic basis. Additionally, individualisation of the SGDs for each participant were limited within each study. Larger trials generally implemented the same device and application across the cohort, with minimal feature matching occurring.

Conclusion

The importance of AAC is clear, yet, there are limited high-powered clinical trials for those with complex communication needs. This review highlighted evidence that supported the use of SGD on communication outcomes, however, studies were often of low quality. There was also a lack of replication of existing studies, meaning that findings could not be further substantiated. Addressing these gaps and expanding on this research is imperative in supporting individuals with complex communication needs.