Li-Ping Yang · Chun-Bo Li · Xiu-Mei Li · Man-Man Zhai · Jing Zhao · Xu-Chu Weng

Abstract Background Developmental dyslexia (DD) is a specific impairment during the acquisition of reading skills and may have a lifelong negative impact on individuals. Reliable estimates of the prevalence of DD serve as the basis for evidence-based health resource allocation and policy making. However, the prevalence of DD in primary school children varies largely across studies. Moreover, it is unclear whether there are differences in prevalence in different genders and writing systems. Hence, the present study aims to conduct a systematic review and meta-analysis to assess the global prevalence of DD and to explore related factors. Methods We will undertake a comprehensive literature search in 14 databases, including EMBASE, PubMed, Web of Science, China National Knowledge Infrastructure and Cochrane, from their inception to June 2021. Cross-sectional and longitudinal studies that describe the prevalence of DD will be eligible. The quality of the included observational studies will be assessed using the Strengthening the Reporting of Observational Studies in Epidemiology statement. The risk of bias will be determined by sensitivity analysis to identify publication bias. Results One meta-analysis will be conducted to estimate the prevalence of DD in primary school children. Heterogeneity will be assessed in terms of the properties of subjects (e.g., gender, grade and writing system) and method of diagnosis in the included primary studies. Subgroup analyses will also be performed for population and secondary outcomes. Conclusion The results will synthesize the prevalence of DD and provide information for policy-makers and public health specialists.

Keywords Developmental dyslexia · Prevalence · Primary school children · Protocol

Introduction

Developmental dyslexia (DD) is a specific impairment characterized by severe and persistent problems in acquiring reading skills that are not solely caused by mental age, visual acuity problems, or inadequate schooling [ 1, 2]. DD, also referred to as specific reading disability or specific reading disorder, is a frequent learning disorder that affects many children. It has been reported that a considerable portion of dyslexic children have academic and social problems [ 3]. Importantly, due to reading failure, they often have higher reading anxiety [ 4– 6], lower positive well-being [ 7], and exhibit learned helplessness and emotion-centered defense mechanisms [ 5]. Therefore, DD is an important public health issue that deserves attention.

Typically, children begin to be formally taught to read after entering primary school. Children’s word reading ability reaches an adult-like level at the end of primary school [ 8]. Formal diagnosis of DD is also usually achieved at the end of the 2nd grade [ 9, 10]. The primary school is, therefore, an important stage at which dyslexic children could be formally diagnosed and intervened [ 11, 12].

It is generally believed that DD affects approximately 5%–12% of children [ 13, 14], and the variability in prevalence may be related to several variables. First, different cutoffcriteria may result in different prevalences. A recent review suggests that the prevalence depends on the severity of the reading problem, with lower rates for more severe problems [ 15]. The common sets of the cutofffor reading achievement are 1 and 1.5 standard deviations (SD) below the mean for the same age [ 9, 15, 16]. Second, environmental variables (e.g., regions and socioeconomic status) and other factors (e.g., grade, subdeficit, sampling method) may also be related to different prevalences. In terms of grade, some studies suggest that prevalence is similar across all grades [ 17– 19], but others have reported mixed results [ 20– 22].

Finally, it is particularly interesting to ask whether and how orthographic depth affects the prevalence of DD. On the one hand, logographic scripts may yield different prevalence estimates compared to alphabetic scripts. In alphabetic scripts in which letters represent phonemes, the reported DD prevalence ranges from 2.28% to 12.70% [ 23, 24], even as high as 15% and 19.90% [ 25, 26]. Unlike alphabetic scripts, logographic scripts such as Chinese have special language characteristics: (1) the smallest written units are characters representing monosyllabic morphemes and (2) grapheme to phoneme mappings are arbitrary [ 27– 29]. Because logographic scripts such as Chinese require rote memorization of characters, it was previously thought to have little difficulty in reading until 1982. Stevenson et al. reported for the first time that DD did exist between Chinese and Japanese [ 30, 31]. On the other hand, scripts have different orthographic depths even within alphabetic writing systems. According to the orthographic depth hypothesis, shallower orthographies are easier to learn than deeper ones [ 32]. It is easier for children to learn how to map letters to known phonological forms in shallower orthography, where the units of written language reliably correspond to the units of spoken language. In contrast, the other two theories (psycholinguistic grain size theory and the grapholinguistic equilibrium hypothesis) propose that DD prevalence is similar in consistent and inconsistent orthography, but its manifestations may be different [ 33, 34].

In addition, the gender ratio is a subject of ongoing debate [ 22, 35]. Most studies reported that more boys suffered from DD than girls, with a gender ratio of approximately 3:1 [ 20, 36, 37], but some studies found no differences in the prevalence of DD between boys and girls [ 16, 22]. The latter interpretation of the overrepresentation of boys in DD prevalence is a result of biased behavioral observation [ 38].

Taken together, numerous previous studies have assessed the prevalence of DD in children, but the results are greatly inconsistent. More importantly, despite the large number and wide distribution of Chinese character users, previous review articles did not well discuss the prevalence of Chinese DD [ 39]. Therefore, it is necessary to include Chinese for meta-analysis. To our knowledge, there is no previous review that has synthesized worldwide evidence from the literature to estimate DD prevalence. The present study thus aims to conduct a systematic review and meta-analysis of previous studies on DD in primary school children. More specifically, the present study aims to address two issues: (1) the prevalence of childhood DD worldwide and (2) whether the prevalence of DD varies by gender, writing system, subdeficit type, cutoffcriteria, grade, sampling method, economic development level and medical and health level.

The present paper reports the protocol of a systematic review and meta-analysis on the prevalence of DD in primary school children. It becomes increasingly common to publish a protocol before a meta-analysis is conducted [ 40– 45]. As a research design, the protocol can provide a research framework for future meta-analyses. It will also be beneficial to make the meta-analysis more transparent and improve its quality and validity.

Methods

Review questions

Two research questions were identified to guide the present systematic review: (1) What is the prevalence of DD in primary school children? (2) Whether the prevalence of DD is infl uenced by gender, writing system, subdeficit type, cutoffcriteria, grade, sampling method, economic development level and medical and health level.

Search strategies

This protocol was registered in PROSPERO (registration number: CRD42021232958), which aims to complete an original systematic review using a search strategy to identify relevant preliminary studies that have been published before June 2021 and indexed in electronic databases, including EMBASE, PubMed, PsycINFO, Web of Science, EBSCO host, ProQuest, the Open Access Thesis and Dissertations database, SpringerLink, Linguistics and Language Behavioral Abstracts, Chongqing VIP database, Wanfang database, Chinese hospital knowledge database, China National Knowledge Infrastructure, and the Cochrane Library. There were no time or language restrictions. The lists of references of related studies will also be manually searched to include all relevant studies.

The search strategy used to retrieve relevant research will be sensitive. Various forms of words, synonyms, plural words and truncations for “developmental dyslexia”, “prevalence”, etc., will be used in different combinations. Supplementary Table 1 reports the search strategy planned for PubMed. We plan to repeatedly modify the search strategy to better adapt to each database until an optimized search strategy of each database is determined. To make the process transparent and repeatable, we plan to put all search strategies in the meta-analysis article. All studies will be imported to Endnote, and duplicates will be identified and deleted.

Inclusion and exclusion criteria

Studies that were conducted to investigate the prevalence of DD in primary school children will be mainly considered for this review. As the primary focus of this study will be on estimating DD prevalence, studies that reported the number of students involved and the number of detections will also be included in the current review for more comprehensive inclusion. The present systematic review will include crosssectional and longitudinal studies.

Inclusion criteria based on PICOs (Cochrane standard) are as follows: (1) P (participants or population): studies with dyslexic children as the study population. The children in the study should have been in primary schools. Because the diagnostic materials and criteria for DD vary greatly, an inclusive approach was used in the literature retrieval. Studies in DD may include individuals whose spelling or reading scores were below the 25th percentile or at least one grade, 1 year, or one standard deviation below the control group [ 46– 48]. All included studies should be conducted in a normal environment with adequate schooling. Studies carried out in prisons, isolated islands, or other inadequate educational environments will be excluded. Any studies without an IQ test will also be excluded; (2) I (intervention) and C (comparison) are not relevant; and (3) O (outcome): estimations of DD prevalence among primary school students; studies involving the number of students involved and the number of detections to facilitate the calculation of detection rates will also be included; S (study design): this review will contain cross-sectional studies and longitudinal studies. The main focus of this study will be to estimate the prevalence of DD. Only studies that reported DD prevalence and that can be calculated to evaluate DD prevalence will be included in the review.

Studies on nonprimary school DD population(s), qualitative studies, clinical studies and secondary analysis studies will be excluded. In addition, studies in which DD is a secondary symptom of a different disorder (for example, attention deficit and hyperactivity disorder or other psychiatric diagnoses) will be excluded from the study. We will not include studies of specific subpopulations of participants (e.g., participants with acute or chronic disease).

Study selection

All studies will be imported to Endnote software to remove duplicates. Next, two review authors (LXM and ZMM) will independently conduct preliminary screening of studies by reading titles and abstracts and then download the full texts of potentially relevant articles for the second round of screening. They will check the screening results and discuss the differences. If no agreement is reached, the final decision will be made by a third reviewer. The above inclusion criteria will be used to select studies for the analysis.

Quality assessment

Two review authors (LXM and ZMM) will independently assess the methodological quality (risk of bias) within each included study according to the STROBE statement (Strengthening the Reporting of Observational Studies in Epidemiology) [ 49]. The STROBE statement contains 22 items and assesses the risk of bias across several domains: the article’s title and abstract, introduction, methods, the results, discussion sections and other information. We will fill the STROBE statement table for all included studies. We will resolve any differences through discussion or by involving a third reviewer.

Data extraction and management

Two authors (LXM and ZMM) will independently perform data extraction using a pretested data extraction form. We will resolve any disagreements about data extraction by discussion or by referring to the original articles. When relevant data are not clear or are unreported, we will contact the corresponding author of the publication for further details. We will enter the extracted data into Review Manager V.5.4.1 (RevMan V.5.4.1) software.

For each study, the basic characteristics (i.e., first author's name; public year; study sites; source of sample, sampling method, sample size; breakdown of sample by gender, and urban versus rural residence; method of screening and diagnosis, dyslexia severity, writing system, socioeconomic status, and the reported prevalence of dyslexia or DD) will be recorded in a table named characteristics of included articles. In the methods of screening and diagnosis, we plan to classify the included studies according to DD subtypes (reading accuracy, reading fl uency, reading skill, reading comprehension, combined). For dyslexia severity, according to different SDs, we intend to divide literatures into four categories: 1 SD, 1.5 SD, 2 SD and without reporting SD. These studies will be designated as high-, middle- and lowincome countries according to the World Bank criteria and as African region, region of the Americas, Southeast Asia region, European region, Eastern Mediterranean region, and Western Pacific region according to the World Health Organization criteria.

Data from articles reporting the results will be combined into one study. Articles other than Chinese or English will be translated into English. The above process will apply to any of these papers.

Results

Analytical method and heterogeneity assessment

After data extraction, the results included in the papers will be categorized and summarized in descriptive tables. The prevalence reported in each study will be stabilized using the Freeman–Tukey double arc-sine transformation prior to computing the pooled prevalence to avoid excessive differences between the prevalence of the original data [ 50, 51]. Then, a forest plot with DD prevalence estimates will be presented. Publication bias will be explored by funnel plots, Begg’s rank method or Egger’s test. Heterogeneity will be analyzed by looking at participants' characteristics, diagnostic methods and case definitions in the included studies. In addition, we will use theI 2statistics (the percentage of variance due to heterogeneity) to select between fixed effects and random effects meta-analyses. AnI2 of ＞ 50% will be considered substantial heterogeneity, in which case the random-effects model will be used; otherwise, the fixedeffects model will be used. In addition, subgroup analyses on writing systems, orthography depth, gender and other variables (e.g., cutoffcriteria, grade, sampling method, subdeficit type, IQ test type, economic development level and medical and health level) will be performed if possible. Fourth, to assess the associations among various sample characteristics and the prevalence of DD, we plan to first conduct univariable meta-regression, if possible, followed by a multivariable meta-regression. The Meta and metafor packages in R 4.1.0 statistical software will be used for this meta-analysis. In this meta-analysis, we will analyze the status of prevalence based on different writing systems through sensitivity analyses.

Patient and public involvement statement

Patients and the public did not participate in the formulation of this protocol. The results will be published in peer-reviewed publications or disseminated in relevant conferences.

Discussion

To date, some meta-analyses on dyslexia or specific reading difficulty have been conducted, which mostly focus on the underlying mechanism of this disorder, such as the auditory processing deficit [ 46], perceptual anchoring deficits [ 47], and procedural learning deficit [ 48]. The planned study is a systematic review and meta-analysis to estimate a worldwide DD prevalence and examine related factors for the first time. The major strength of this work is that a large number and range of children will be included to estimate the prevalence of dyslexia. We would also like to note a few limitations of the present study. First, the children who we would like to examine come from normal schools. We suppose that all children have an adequate educational environment, which is stated in previous studies and meta-analyses [ 52]. Although this assumption is applicable to most cases, it is not entirely appropriate. Second, although we plan to search more than 10 databases, it is still possible that we may not be able to include all studies on the prevalence of DD. Third, the publications that report the subtypes of this deficit and other variables may not be sufficient. Therefore, subgroup analyses are not guaranteed to be performed in some cases.

In summary, this review study will synthesize worldwide evidence and provide an estimation of DD prevalence in primary school children. The findings of this review will be of interest to teachers, psychologists, policy makers and others working with primary school students. Moreover, the analysis of different subgroups can provide more evidence for revealing the infl uencing factors of dyslexia and in turn contribute to understanding the mechanism underlying this disorder.

Supplementary InformationThe online version contains supplementary material available at https:// doi. org/ 10. 1007/ s12519- 022- 00572-y.

Author contributionsYLP conceived and designed the protocol, and drafted the protocol manuscript. WXC and LCB critically revised the manuscript for methodological and intellectual content. LXM and ZMM participated in the development of the search strategy. ZJ conceived and designed the protocol, and critically revised the manuscript for methodological and intellectual content. All the authors approved the final version of the manuscript.

FundingThe study was supported by the Key-Area Research and Development Program of Guangdong Province (No. 2019B030335001), the National Science Foundation of China (No. 20 & ZD296; No. 32171063), the Science and Technology Project of Guangzhou City (No. 201804020085) and Shanghai Clinical Research Center for Mental Health (No. 19MC1911100).

Data availabilityData sharing not applicable to this article as no data sets were generated or analysed during the current study.

Declarations

Ethical approvalEthical approvalis not applicable in this study.

Conflict of interestThe authors declare no confl ict of interest.