Reading comprehension intervention in populations with developmental language disorder: A scoping review · Lam JHY, Wang J, Pratt AS et al.

Method

The current study adopted the scoping review framework proposed by Arksey and O’Malley (2005). The framework includes five stages of the review process, including (1) identification of the research question, (2) identification of relevant studies, (3) selection of relevant studies, (4) charting of findings, and (5) reporting of results. This framework was chosen given that it can illustrate the field’s interest in a rigorous and transparent manner, identify gaps in the literature and summarize the research findings (Arksey & O’Malley, 2005). Furthermore, this framework has been adopted in the field of speech-language pathology related to clinical areas with limited research (e.g., Tucci & Choi, 2023).

Identification of research question

The current review was motivated by the following research question: What is known from the existing literature about reading comprehension interventions in populations with DLD? In order to answer that question, we summarize the existing evidence base, as well as gaps and limitations in the existing literature. Though this question requires the synthesis of the evidence regarding efficacy and effectiveness, our primary purpose was to map the breadth, characteristics, and limitations in the literature on reading comprehension interventions for children with DLD. Our methodology aligns with the definition of a broad-scope review (Chandler et al., 2019) and seeks to include all available evidence—regardless of differences in designs, outcomes, and reporting style—in order to describe the existing literature and identify areas for future research, which might later support a systematic or meta-analytic review.

The present scoping review is pre-registered on the Open Science Framework Registries (Name: Reading comprehension intervention in children with DLD: A scoping review; Registration number: https://osf.io/mnc7t/).

Identification of relevant studies

Relevant studies were located in three electronic databases, including ERIC, EBSCO, and PsycINFO. The literature search on these databases was performed on Aug 7, 2022, and included all relevant studies through this date. The search strings combined terms for intervention, training, or treatment; language disorder, impairment, or disability; and reading comprehension. Complete search strings are available in Supplementary Material 1. Results of the literature search were then exported as RIS files and imported into Covidence systematic review software for screening and extraction (Covidence, 2023).

The following five criteria were applied for the eligibility criteria for inclusion in the current scoping review: (1) included a DLD or language-impaired group, which can be identified by SLPs at research, clinical or school settings, standardized assessments, or from any formal clinical records; (2) included a treatment or intervention component; (3) included at least one reading comprehension measure as an outcome or generalization measure in post-intervention or follow-up assessment; (4) published in a peer-reviewed journal and included original empirical findings; (5) published in English.

Study selection

The initial search generated 4867 references, which included 968 duplicates. 3894 unique references were imported to Covidence for the screening and extraction processes. The first author performed the title and abstract screening for all references, while the second author performed a second rater reliability check on 20 % of the references. The interrater reliability was 96.41 % (753/781) and inconsistency was resolved by the third author.

A subset of 248 articles was further screened at the full-text level using the criteria mentioned above. The first author screened all full text articles, while the second author performed a second rater reliability check on 20 % of them. Interrater reliability was 96 % (48/50). Inconsistencies were resolved by the third author. After reviewing all the articles, 24 articles were identified that met the inclusion criteria. Reference lists of the included articles were screened, and no additional articles were identified. Thus, 24 studies were ultimately included in the current systematic review. Fig. 2 shows the search and selection process.

Charting of findings

Study information was extracted for synthesis by the first author on all 24 included articles into a Google Sheet. There were three major areas for data extraction: participant characteristics, intervention characteristics, and intervention outcomes. The participant characteristics extracted included the sample size, sex, age (both mean and range), race, monolingual/bilingual status, and language(s) spoken by the participants. In addition, the terminology and the identification process of the DLD group were also recorded.

Regarding the intervention characteristics, duration, interventionist, setting, dose frequency, dosage, intensity, service delivery, participant size, social validity, and fidelity of the intervention were coded. Moreover, the description of the intervention was extracted together with language skills targeted in the intervention. Coding of language skills was based on the AVR (Duke & Cartwright, 2021). Four domains were included in this model: active self-regulation, word recognition, bridging process, and language comprehension (specific skills were specified within the four domains as shown in Fig. 1). For interventions that had multiple language domains/skills targeted, the first and second authors read the paper together and decided the primary domains/skills and other targeted language domains/skills were also noted.

The extracted intervention outcomes consist of the reading outcome measures and the statistics reported. The reading measurements adopted in the studies were extracted with the information of the reading text(s), question type(s), answer method(s), and reliability. For the statistics reported, the mean and standard deviation of the reading measurements were extracted at pre-test, post-test, and follow-up assessment if they were reported. The statistical analysis adopted to measure the difference in reading comprehension ability was extracted with the statistical test(s) conducted, statistical coefficient(s), and the significance level. The second author randomly completed the extraction 25 % of the included articles (6/24) and interrater agreement was 92.07 % (279/303). Discrepancies were resolved before coders moved into independent coding. The complete dataset is available in the replication materials (https://osf.io/5vs7n).

Intervention biases

The source of bias in literature reviews has received increasing attention. Thus, it is important to consider the risk of bias in the included studies rather than only summarizing the characteristics in the synthesis (Page et al., 2021). To assess the risk of bias, Quality Assessment of Controlled Intervention Studies, Quality Assessment Tool for Before-After (Pre-Post) Studies With No Control Group, and Quality Assessment Tool for Case Series Studies were adopted to assess the risk of bias for included studies based on the types of intervention design (National Institutes of Health, 2015). The first author completed the coding for all included studies, while the second author performed a second rater reliability check on 25 % of the included articles (6/24). The exact agreement for the risk of bias assessment was 95.71 % (67/70).