Beginning reading interventions for children and adolescents with intellectual disability

  • Protocol
  • Intervention

Authors


Abstract

This is the protocol for a review and there is no abstract. The objectives are as follows:

To assess the effectiveness of interventions for teaching beginning reading skills to children and adolescents with intellectual disability.

Background

Description of the condition

The International Classification of Diseases describes intellectual disability as "a condition of arrested or incomplete development of the mind, which is especially characterized by impairment of skills manifested during the developmental period, skills which contribute to the overall level of intelligence (i.e., cognitive, language, motor, and social abilities)" (WHO 2010). Intellectual disability is "characterized by significant limitations both in intellectual functioning (reasoning, learning, problem solving) and in adaptive behavior, which covers a range of everyday social and practical skills….[and] originates before the age of 18" (p.1, AAIDD 2011). Significant limitation is most commonly defined as a score (with measurement error considered) on a standardized, norm-referenced assessment that is approximately two standard deviations or more below the population mean (e.g., an IQ of 70 or less; Luckasson 2002). Other definitions have been adopted by various organizations (e.g., Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM-IV; APA 2000); International Classification of Functioning, Disability and Health (ICF) (WHO 2001)), although most emphasize three dimensions including intellectual functioning, adaptive behavior, and early manifestation (Polloway 2011).

Globally, an array of terms are used to describe the condition of intellectual disability. Across 147 countries, the most commonly used terms include mental retardation (76.0%), intellectual disabilities (56.8%), mental handicap (39.7%), learning disabilities (32.3%), developmental disabilities (22.6%), mental deficiency (17.2%), and mental subnormality (11.6%) (WHO 2007). However, many professional and governmental organizations and the broader professional community have shifted to the use of intellectual disability (Schalock 2011), and advocates remain focused on increasing the consistent use of the term intellectual disability by governmental and professional organizations worldwide. Schalock 2007 provided the rationale for the change:

"The term intellectual disability (a) reflects the changed construct of disability proposed by AAIDD [American Association of Intellectual and Developmental Disabilities] and WHO [World Health Organization]; (b) aligns better with current professional practices that are focused on functional behaviors and contextual factors; (c) provides a logical basis for individualized supports provision due to its basis in a social-ecological framework; (d) is less offensive to persons with disabilities; and (e) is more consistent with international terminology (p.120, Schalock 2007)."

The prevalence of intellectual disability is estimated to be 10.37 in 1000, globally (Maulik 2011). Maulik 2011 reported variation in prevalence rates across studies, with higher incidences of intellectual disability found when samples were drawn from low- and middle-income countries, rural and urban slum populations, and child or adolescent populations. The overall prevalence rate is, however, similar to previous estimates that approximately 1% of students from high-income countries are identified as having intellectual disability (Polloway 2010). Although the prevalence rate is low, it is worth noting that this represents a substantial number of individuals worldwide who are in need of educational services - approximately 31 million children, adolescents, and young adults up to 24 years of age (United States Census Bureau 2013).

Causation is often unknown, particularly for individuals with intellectual disability who have higher IQ scores (i.e., slightly below the cut-off of two standard deviations below the mean; Snell 2009). Cases with an identified cause result in various levels of intellectual disability but are most often associated with more significant cognitive impairment (Polloway 2006). Known causes include specifiable biological causes such as chromosomal abnormalities (e.g., Down syndrome, fragile X syndrome), metabolic disorders (e.g., phenylketonuria), and postnatal brain disease (e.g., neurofibromatosis). Additionally, early infection (e.g., congenital rubella), trauma (e.g., oxygen deprivation during birth, severe head injury), and exposure to toxins (e.g., lead poisoning) are causally connected to intellectual disability. Poverty and cultural deprivation have also been associated with an increased risk of intellectual disability, possibly due to deficits in nutrition, education, and health care, and increased environmental risks (Hardman 2014). However, the relationship between biological and environmental causes of intellectual disability is complex. For many individuals, the precise contribution of possible causes cannot be confirmed (Polloway 2011). Additional research is needed to understand better the interactions between environmental and biological causes (Dykens 1995; Tunnicliffe 2011).

Quality of life of individuals with intellectual disability has improved in many countries since the last half of the 20th century (Weymeyer 2013). This improvement is due, in part, to changes in how societies view individuals with intellectual disability (Spooner 2011). Representative of this changing view is the number of individuals with intellectual disability who live in large institutions. Beadle-Brown and colleagues conducted a review of the literature and found that many countries (e.g., Canada, England, Norway, Sweden, United States) have substantially decreased the numbers of individuals with intellectual disability who live in institutionalized settings during this time (Beadle-Brown 2007). Other countries (e.g., Belgium, Germany, Greece, Spain, The Netherlands) have only more recently begun the process of deinstitutionalization and institutional care continues to be the most common living condition for individuals with intellectual disability. Legal protections for individuals with intellectual disability have also changed since the 1950s in many countries (Weymeyer 2013). The World Health Organization surveyed 147 participating countries and documented that 59.2% had a national policy or program specifically related to intellectual disability, and that 22.4% referred to intellectual disability in other policies (WHO 2007). However, 18.4% of respondents indicated that intellectual disability was not covered by any policy or program. In terms of education, 77.8% of countries addressed intellectual disability in a specific education policy or program. These improvements are not, however, uniform across countries. A small number of countries (e.g., Taiwan) continue to advocate for increased institutionalization (Beadle-Brown 2007). In addition, the level of change in living conditions varies greatly within and across countries, largely dependent on support from local and national organizations and the availability of resources (IASSID 2001). For example, lower-income countries were less likely than high-income countries to address intellectual disability in their education policies or programs (66.7% versus 90.5%, respectively).

There is high variability in adult outcomes for individuals with intellectual disability, often moderated by the level of cognitive impairment and the availability of resources to provide intervention and ongoing support (Stein 2011). Levels of employment range from 37% to 70% of adults with intellectual disability; however, individuals work, on average, only 20 to 28 hours per week at wages substantially below median income (Huang 2010). Intellectual disability is also associated with an increased risk for physical (e.g., obesity, mobility impairments) and mental (e.g., psychiatric disorders, anxiety disorders) health problems (Stein 2011). Further, adults with intellectual disability often have smaller social networks whose members are most often family members, care providers, or disabled peers (Huang 2010). Additional research is needed to document how ongoing changes in how individuals with intellectual disability are viewed by society have effected adult outcomes.

It is not surprising that individuals with intellectual disability are less skilled readers than their typically developing peers - reading is a complex cognitive skill and poorer performance on cognitive tasks is a defining characteristic of intellectual disability (Stanovich 1985). Reading difficulty has been reported as the most common secondary condition that causes significant limitation for individuals with intellectual disability (Koritsas 2011). As would be expected, individuals with intellectual disability consistently underperform compared to typically developing peers who are matched on verbal ability on key reading skills (e.g., word recognition, phonological decoding; Channell 2013). Wei 2011 demonstrated that reading skills of students with intellectual disability are also substantially poorer than peers with other disabilities (e.g., autism, learning disabilities). The authors estimated longitudinal growth trajectories on letter-word identification and passage comprehension for 3421 students with various disabilities across the ages of 7 to 17 years. At all points in time, the average performance of students with intellectual disability was lower than that of students in all other disability categories. Katims 2000 examined the reading skills of a group of 132 elementary, middle-, and high-school students with intellectual disability (IQs between 29 and 76). The author reported that only 22.2% had obtained minimum literacy, defined as (a) reading and comprehending connected text at the primer level, (b) being able to write at least two graphemes representing spoken sounds, and (c) being able to write at least two words correctly. Further, Lemons and colleagues estimated the reading skill of a sample of 7440 students with disabilities using curriculum-based measurement (CBM; Deno 1985) (Lemons 2013). In this sample, only 1.9% and 3.5% of students with intellectual disability in grades eight (n = 663) and 11 (n = 498), respectively, were able to read a passage at a fifth grade level.

Despite evidence of substantial impairment in reading ability for persons with intellectual disability, individual differences in reading skill have also been documented. Severity of intellectual impairment has frequently been cited as a significant predictor of variation in academic performance for individuals with intellectual disability. For example, Turner 2008 examined longitudinal predictors of academic attainment in a sample of 71 individuals with intellectual disability (i.e., Down syndrome). Mental age assessed at time one (mean age nine years) accounted for the largest portion of variance in academic attainment (i.e., a composite of reading, writing, and numeracy skill) assessed at time three (mean age 21 years). Additionally, researchers have explored relationships between cognitive ability and reading skill for a variety of alphabetic languages (i.e., languages in which phonemes (individual spoken sounds) are represented by graphemes (written symbols used to represent the sounds)). Across an array of languages, intelligence is statistically significantly correlated with a range of readings skills for individuals with intellectual disability (Cohen 2001 (French); Levy 2011 (Hebrew); Soltani 2013 (Persian)).

Description of the intervention

Literacy, broadly defined, encompasses "using printed and written information to function in society, to achieve one's goals, and to develop one's knowledge and potential" (Kirsch 2002, p.2). Beginning reading instruction focuses on teaching students to identify printed letters and words. The process of learning to read entails "coming to know how one's writing system works" (Perfetti 1998, p.6). In other words, reading involves learning how spoken language is encoded in the printed form of the writing system. This learning is a necessary component in developing higher-level reading skills, including reading comprehension and being able to get meaning from print. The focus of this review will be on beginning reading interventions that target initial acquisition of reading skills, not on interventions targeting higher-level reading skills such as reading comprehension, reading vocabulary, and oral reading fluency.

For individuals with intellectual disability, two approaches, sight word and phonics, have traditionally been used to teach beginning reading skills to children and adolescents with intellectual disability.

Sight word approach

The most common approach for teaching students with intellectual disability has been a sight word approach by which students memorize a set of words (Browder 1998). Often, words are selected for their functional use (e.g., words that facilitate grocery shopping or safety), and a visual (e.g., photograph, icon) or physical (e.g., real or artificial object) representation may be paired with the printed word to facilitate learning. Students with intellectual disability who have been provided systematic sight word instruction have learned to identify a fairly large number of words (Browder 2006). This approach has traditionally been preferred over phonics-based approaches (described below), based on the presumption that attempting to teach individuals with intellectual disability to read by understanding relationships between phonemes and graphemes would be too cognitively taxing and thus unsuccessful (Browder 2009). Despite its effectiveness at increasing the ability to read a set number of taught words, a sight word approach is limited. Individuals who learn to read via sight word instruction have only a limited understanding of how the writing system encodes spoken language (i.e., focusing on whole words rather than graphemes limits a learner's ability to benefit from the connection between print and sounds). Thus, students do not learn how to generalize their learning beyond directly taught words. More specifically, students do not develop the ability to read novel, non-taught words - a critical component of literacy acquisition.

Phonics-based approach

Due to the limitations of a sight word approach, several researchers have begun to explore phonics-based approaches to reading intervention for individuals with intellectual disability (Allor 2010; Lemons 2010; Browder 2012). These approaches focus on teaching students to identify and manipulate phonemes in spoken language and to connect these with printed graphemes. Two core components include: (a) phonological awareness, and (b) phonics. Phonological awareness is the ability to hear and manipulate sounds in spoken language (Carnine 2010). The continuum of skills included under the broader category includes the ability to identify rhyme and alliteration, to segment a spoken sentence into words, to blend and segment syllables, to manipulate onset and rime, and to segment, blend, and manipulate individual phonemes in words. Phonics involves teaching students to make connections between the sounds in spoken language and the letters used to represent the sounds in print (Hougen 2012). Instruction involves teaching students letter sounds, decoding (converting print to spoken language), and encoding (converting spoken language into print).

Perfetti 1998 explained how learning a writing system occurs across different languages. The authors specify that for alphabetic languages (e.g., English, Italian, Russian), the key principle to be learned is the alphabetic principle - an understanding of the association of meaningless printed units (i.e., letters or graphemes) to meaningless units of the spoken language (individual sounds or phonemes). In modified alphabetic languages (e.g., Hebrew, Egyptian), print is used to represent consonants, but not often vowels. The authors further explain how the process is similar for almost all languages with some variation in the 'size' of the spoken language unit that is associated with the printed graphic. For example, Japanese relies on the syllabic principle whereby graphic units are associated with spoken syllables; Chinese is based on a morphological principle in which the written units encode morphemes. (Chinese is additionally complex in that graphic symbols also encode syllables, thus identifying it as morphosyllabic language may be more appropriate (see Perfetti 1998; Perfetti 2003)). Regardless of language, the critical aspect of learning to read is to recognize that written language represents speech and to understand the relationship between the two.

Phonics-based approaches to teaching children to read in English are supported by three decades of research (Adams 1990; Snow 1998; National Reading Panel 2000). Due to the strong empirical support, this type of reading instruction is mandated in Australia (Rowe 2005), the United Kingdom (Rose 2009), and the United States (No Child Left Behind Act of 2002). However, individuals with intellectual disability were predominately excluded from empirical evaluations of phonics-based approaches. Further, a recently conducted meta-analysis evaluating phonics instruction for English-speaking poor readers was unable to draw conclusions about the efficacy of phonics instruction for individuals with intellectual disability due to the limited number of individuals with intellectual disability included in the reviewed studies (McArthur 2012). Thus, it remains unclear how effective phonics-based approaches are for individuals with intellectual disability.

Syntheses on reading instruction for individuals with intellectual disability

Browder and colleagues have completed the most recent meta-analytic review on teaching individuals with intellectual disability to read in English (Browder 2006). The authors identified 128 studies (88 single subject design; 40 group design) involving 1123 participants. Studies predominantly targeted acquisition of functional sight words; fewer examined phonological awareness (n = 5) or phonics instruction (n = 13). Following an evaluation of study quality based on previously published indicators (Gersten 2005; Horner 2005), the authors concluded that the evidence base supports the teaching of sight words using systematic prompting techniques in a repeated trial format. The authors were unable to identify a sufficient number of high quality studies to verify other evidence-based approaches for teaching reading to students with intellectual disability (e.g., phonological awareness, phonics). In the most recently published meta-analysis, focused on phonics-based approaches for individuals with intellectual disability, the authors identified seven studies (Joseph 2004). Findings from the studies indicate that students with intellectual disability were able to benefit from phonics-based instruction. However, many of the studies were focused on limited aspects of phonics (e.g., classifying letters as consonants or vowels) and the authors did not evaluate studies for rigor of experimental design.

How the intervention might work

We presume that interventions focused on the targeted components of reading will work in much the same manner for students with learning disabilities as they do for students who are typically developing. Instruction assists learners in (a) understanding that spoken language consists of units (e.g., words, syllables, and phonemes) that can be broken apart and manipulated, and (b) associating these spoken units with the graphic units used to represent them in speech. A student is most often first exposed to a new skill (e.g., learning the sound of the letter 't') with a model by an instructor. The instructor then provides multiple opportunities for the student to practice the skill while receiving immediate, corrective feedback. Instructor support or scaffolding is purposefully decreased until the student is able to perform the task independently. Then, the student is provided with multiple opportunities to practice the skill independently and to work on improving automaticity or fluency. Various instructional strategies have been demonstrated to be effective in accomplishing these learning objectives (direct, explicit instruction, Allor 2010; systematic prompting and fading, Browder 2006). Further, emergent language skills, which are often delayed in individuals with intellectual disability, are highly predictive of response to early reading intervention for individuals with (Steele 2013) and without intellectual disability (Scarborough 1990). Thus, as proposed by Burgoyne 2009, interventions that incorporate an oral language component may be more effective at increasing early reading skills than interventions that do not.

We expect that the rate of learning may be slower for students with intellectual disability compared to other types of learners and that instruction may need to be provided for a longer period of time (e.g., multiple school years) and/or at a greater level of intensity (e.g., one-on-one instruction, additional minutes per week). We also believe that several important features may moderate learning. First, as demonstrated in previous studies (Snell 2009), we expect student outcomes to be associated with severity of impairment. Higher IQ scores will likely be associated with better intervention effects. Second, as learning to read involves developing an understanding of how spoken language is represented in print, we would hypothesize that response to reading intervention will be influenced by language skills (e.g., receptive vocabulary). Third, type of instruction (sight word, phonics) may influence outcomes. Studies that focus on acquisition of functional sight words will likely demonstrate stronger effects in sight word learning, whereas interventions targeting phonics skills (e.g., letter sound knowledge) would be presumed to have stronger effects on targeted skills. Fourth, increases in the intensity of intervention (e.g., frequency, duration, group size) would be hypothesized to be associated with greater outcomes (Vaughn 2012). Fifth, variations in environment may moderate treatment effects. We aim to code for and, if possible, examine moderator effects for educational setting of the study (e.g., special school, general education setting) and education quality of the country in which the study was conducted as indexed by the United Nation's Education Index (United Nations Development Programme 2013). Sixth, differences in language (alphabetic versus syllabic; opaque versus transparent) may moderate outcomes. Previous work has demonstrated that, for typically developing children, learning to read is easier in alphabetic languages that are more transparent (i.e., consistent sound-symbol correspondences; e.g., Finnish, Hungarian) compared to languages that are more opaque (i.e., less consistency between sound-symbol correspondences; e.g., French, English) (Ellis 2004; Ziegler 2010). We hypothesize that individuals with intellectual disability may also have more success in learning to read when their language is more transparent.

Why it is important to do this review

Historically, many educators assumed that individuals with intellectual disability were not capable of learning to read; thus reading instruction was not considered appropriate (Katims 2000). However, across a number of countries (e.g., Australia, United Kingdom, United States), societal expectations for individuals with intellectual disability have drastically changed in the last 20 years and schools are now expected to teach academic content (e.g., reading, mathematics, science) to students with intellectual disability (Browder 2009).

Previous reviews were limited to English language instruction and conducted between seven and 15 years ago (Browder 1998; Joseph 2004; Browder 2006). Our aim is to conduct a more comprehensive, up-to-date synthesis by examining the effectiveness of reading intervention across languages and instructional methods. This review will allow us to explore the effects of various types of reading intervention and may allow us to examine important moderating factors that may effect student outcomes, including severity of impairment, method of instruction, intensity of instruction, or characteristics of the spoken language.

Our focus includes initial reading acquisition interventions. We will seek to include studies in which one or more elements of reading instruction, including (a) phonological awareness, (b) letter-sound or grapheme-phoneme correspondences, (c) decoding or phonics, (d) encoding or spelling, or (e) reading of words or connected text, was targeted by the intervention and included as a dependent variable. There are no exclusionary criteria related to intervention language or instructional method (e.g., sight word, phonics, other).

Objectives

To assess the effectiveness of interventions for teaching beginning reading skills to children and adolescents with intellectual disability.

Methods

Criteria for considering studies for this review

Types of studies

Randomized and quasi-randomized controlled trials (in which systematic methods of allocation to groups are used, but which are not truly random e.g., day of week, case number presentation).

Types of participants

Children aged 4 to 21 with intellectual disabilities, defined as an IQ equal to or greater than two standard deviations below the population mean, irrespective of language. When a specific IQ is not provided but a descriptive term is used, we will accept studies that describe the sample as having 'intellectual disability' or 'mental retardation', which are the two most common terms used to refer to intellectual disabilities according to the WHO Intellectual Disability Atlas (WHO 2007).

Types of interventions

Literacy interventions, irrespective of instructional method (e.g., sight word, phonics, other), comprising one or more elements of reading acquisition, including:

  • phonological awareness;

  • letter-sound or grapheme-phoneme correspondences;

  • decoding or phonics;

  • encoding or spelling;

  • sight word learning;

  • word recognition;

  • storybook reading; and

  • text reading.

Control intervention: no treatment control, wait-list control, treatment as usual, attention control, or alternate non-reading instruction control (e.g., mathematics instruction).

Excluded: other treatment control designs, i.e., where only two reading interventions are compared one to another.

Types of outcome measures

Primary outcomes
  1. Phonological awareness (e.g., ability to blend individual phonemes into a word).

  2. Sound-symbol correspondence (e.g., ability to produce phonemes correctly when presented with isolated graphemes).

  3. Word reading (e.g., ability to pronounce a written word correctly).

  4. Decoding (e.g., ability to use sound-symbol knowledge to decode novel words or pseudo-words).

  5. Adverse events (e.g., problem behavior, social stigmatism).

Secondary outcomes
  1. Supplemental measures of literacy, including vocabulary, comprehension, spelling, writing, and/or oral reading fluency.

  2. Supplemental measures of language skills, treatment acceptability, attitudes towards reading, self efficacy related to reading, and behavioral outcomes, if reported.

Outcomes might be measured using standardized assessments, qualitative data (for example, social validity), parent and/or teacher rating scales, and behavioral observation. Due to the likely variability in quality, we will consider all measures but we will discuss the evidence of their reliability and validity and report this in the 'Characteristics of included studies' table. Where both parent and teacher measures are used, we will prioritize teacher-reported measures.

We will group outcome time points as follows: immediately post-intervention, 1 to 5 months post-intervention, 6 to 11 months post-intervention, 12 to 23 months post-intervention, 24 to 35 months post-intervention, and so on.

We will report the primary and secondary outcomes in the 'Summary of findings' table.

Search methods for identification of studies

Electronic searches

We will search the following electronic databases for all available years, with no date limits or language restrictions.

  1. Cochrane Central Register of Controlled Trials (CENTRAL), part of The Cochrane LIbrary

  2. Ovid MEDLINE

  3. EMBASE

  4. CINAHLPlus

  5. PsycINFO

  6. ERIC

  7. British Education Index

  8. Australian Education Index

  9. Science Citation Index (SCI)

  10. Social Science Citation Index (SSCI)

  11. Conference Proceedings Citation Index – Science

  12. Conference Proceedings Citation Index – Social Sciences and Humanities

  13. Cochrane Database of Systematic Reviews (CDSR), part of The Cochrane LIbrary

  14. Database of Abstracts of Reviews of Effects (DARE), part of The Cochrane LIbrary

  15. WorldCat (limited to theses and dissertations) (worldcat.org/)

  16. EPPI-Centre Database of Education Research (eppi.ioe.ac.uk/webdatabases)

  17. OpenGrey (opengrey.eu/)

  18. National Technical Information Service (NTIS) (ntis.gov/)

  19. ClinicalTrials.gov (clinicaltrials.gov/)

  20. ICTRP (who.int/ictrp/en/)

We will use the following Ovid MEDLINE search strategy, and adapt the search terms and syntax as appropriate for other sources.

1 exp Mental Retardation/
2 exp Intellectual disability/
3 Mentally Disabled Persons/
4 Developmental Disabilities/
5 Learning disorders/
6 ((mental$ or intellectual$) adj2 (impair$ or retard$ or disab$ or defici$ or handicap$ or subnormal$ or sub-normal$)).tw.
7 ((learning$ or cognit$) adj2 (impair$ or retard$ or disab$ or defici$ or difficulty or difficulties or handicap$ or subnormal$ or sub-normal$)).tw.
8 ((profound$ or mild$ or moderate$ or multiple or severe$) adj1 (handicap$ or disab$)).tw.
9 feeble-minded$.tw.
10 moron$.tw.
11 imbecil$.tw.
12 (down$ adj3 syndrome$).tw.
13 (prader adj3 willi).tw.
14 (williams adj3 syndrome).tw.
15 fragile X.tw.
16 ((Martin Bell or FRAXE or FRAXA) adj3 syndrome$).tw.
17 ((low or borderline) adj1 IQ).tw.
18 or/1-17
19 Reading/
20 (read or reading).tw.
21 (literacy adj3 (intervention$ or program$ or instruction$ or skill$ or activit$ or develop$ or strateg$)).tw.
22 Phonetics/
23 (phonetic$ or phonic$ or phonolog$ or phonemic or phoneme$).tw.
24 grapheme$.tw.
25 (word$ adj3 (decod$ or de-cod$ or encod$)).tw.
26 (alliteration or alphabeti#e or alphabeti#ation or rhyme or rhyming).tw.
27 (sound adj1 symbol$).tw.
28 (letter adj1 (name$ or knowledge or sound)).tw.
29 ((word$ or sound$) adj3 (blend$ or connect$ or segment$ or onset$ or rime$)).tw.
30 (sight adj1 (method or vocabulary or word)).tw.
31 word recognition.tw.
32 orthograph$.tw.
33 (rebus or story-map$).tw.
34 or/19-33
35 randomized controlled trial.pt.
36 controlled clinical trial.pt.
37 randomi#ed.ab.
38 placebo$.ab.
39 randomly.ab.
40 trial.ab.
41 groups.ab.
42 or/35-41
43 exp animals/ not humans.sh.
44 42 not 43
45 18 and 34 and 44

Searching other resources

We will check the reference lists of relevant publications in order to find additional studies which may have been missed by the electronic searches. We will contact the authors of eligible studies to ask if they know of other relevant unpublished or ongoing trials. In addition, we will use Google Scholar to search for papers which cite relevant studies, as well as searching for papers on similar topics by the authors of our included studies.

Data collection and analysis

Selection of studies

Two review authors (BR and DM) will independently screen the titles and abstracts yielded by the search against the inclusion criteria listed above. These authors will retrieve and assess the full texts of any papers about which there is uncertainty. We will seek additional information from report authors as necessary to resolve questions about the relevance or methodology of a study. We will resolve disagreements about eligibility through discussion and, when disagreements cannot be resolved, we will seek advice from a mediator (DH). We will record the reasons for excluding trials. Neither of the review authors will be blind to the journal titles or to the study authors and institutions.

Data extraction and management

Two review authors (DH and CL) will independently extract data for each trial using a data extraction form to collect information about the population, intervention, randomization methods, blinding, sample size, outcome measures, follow-up duration, attrition and handling of missing data, and methods of analysis. When data are missing, one author (CL) will contact the authors to request additional information. If further information cannot be obtained, we will code the variables in question as 'missing'.

Assessment of risk of bias in included studies

We will independently assess risk of bias using The Cochrane Collaboration's tool for assessing risk of bias (Higgins 2008a). We will resolve any disagreements by discussion and, if necessary, disagreements will be arbitrated by a third party. We will use the tool to assess the following domains: sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessment, incomplete outcome data, selective outcome reporting, and other potential sources of bias.

We will present the 'Risk of bias' assessments in a table where the judgment of the review authors (low, high, or unclear risk of bias) will be followed by a text box providing details on the available information that led to each judgment. We will use GRADE to assess the risk of bias for the evidence pertaining to each outcome, which we will report by outcome in the 'Summary of findings' table.

Sequence generation

Was the sequence generation method used adequate? We will judge the risk of bias as follows:

  • 'low' when participants were allocated to treatment conditions using randomization such as computer-generated random numbers, a random numbers table, or coin-tossing;

  • 'unclear' when the randomization method was not clearly stated or unknown;

  • 'high' when randomization was not used in accordance with any of the above methods.

Allocation concealment

Was allocation adequately concealed? We will judge the risk of bias as follows:

  • 'low' when participants and researchers were unaware of the participants' future allocation to treatment condition until after decisions about eligibility were made and informed consent was obtained;

  • 'unclear' when allocation concealment was not clearly stated or unknown;

  • 'high' when allocation was not concealed from either participants before informed consent or from researchers before decisions about inclusion were made, or allocation concealment was not used.

Blinding of participants and personnel

Were participants and personnel blind to which participants were in the treatment group? We will judge the risk of bias as follows:

  • 'low' when blinding of participants and key personnel was ensured;

  • 'unclear' when blinding of participants and key personnel was not reported;

  • 'high' when there was no or incomplete blinding of participants and key personnel, or blinding of participants and key personnel was attempted but was likely to have been broken.

Blinding of outcome assessment

Were outcome assessors blind to which participants were in the treatment group? We will judge the risk of bias as follows:

  • 'low' when blinding of outcome assessment was ensured;

  • 'unclear' when there was not adequate information provided in the study report to determine blinding of outcome assessment, or blinding of outcome assessment was not addressed;

  • 'high' when blinding of outcome assessment was not ensured.

Incomplete outcome data

Did the trial authors deal adequately with missing data? We will judge the risk of bias as follows:

  • 'low' when the number of participants randomized to groups is clear and it is clear that all participants completed the trials;

  • 'unclear' when information about which participants completed the study could not be acquired by contacting the researchers of the study;

  • 'high' when there is clear evidence that there was attrition or exclusion from analysis in at least one participant group that is likely related to the true outcome.

Selective outcome reporting

Did the authors of the trial omit to report on any of their outcomes? We will judge the risk of bias as follows:

  • 'low' when it is clear that the published report includes all expected outcomes;

  • 'unclear' when it is not clear whether other data were collected or not reported;

  • 'high' when the data for one or more expected outcomes are missing.

Other potential sources of bias

Through assessment, we will determine whether any other bias is present in the trial, such as changing methods during the trial or other anomalies.

Measures of treatment effect

Dichotomous data

Where dichotomous data are presented, we will calculate an odds ratio (OR) with a 95% confidence interval (CI) for each outcome in each trial (Higgins 2008a).

Continuous data

We will analyze continuous data when means and standard deviations are presented in the study papers, are made available by the authors of the trials, or are calculable from the available data. Where outcomes are measured using the same scale, we will calculate a mean difference to determine the differences in mean scores between groups. Where similar outcomes are measured using different scales, we will calculate a standardized mean difference using Hedges g with small sample correction (Hedges 1985). We anticipate that different measures and scales will be used across the included studies, therefore the meta-analysis will combine all three types of effect sizes by transforming all metrics to Hedges g.

Multiple outcomes

If a study includes two functionally equivalent measures, we will estimate a mean effect size if the mean and standard deviation for both measures are included, provided the correlation between the outcomes is provided. If the correlation is missing, we will contact the authors to obtain the correlation first and, if unsuccessful, we will examine assessment protocols to determine whether a correlation can be located for estimating a mean effect size. If the correlation between outcomes cannot be obtained or if more than two functionally equivalent measures are included, we will conduct a sensitivity analysis by including each functionally equivalent outcome stepwise to determine the magnitude of effect on the overall synthesis by including the distinct outcomes.

Unit of analysis issues

We expect studies in this area to randomize individual participants to one or more study groups. If we identify cluster-randomized trials (in which groups of participants are randomized e.g., classes, schools, or cross-over designs) in which participants are randomly allocated to study arms consisting of a sequence of two or more treatments given consecutively (Higgins 2008b), we will take the following steps to ensure that we avoid unit of analysis errors.

Cluster-randomized trials

We will adhere to the advice provided on statistical methods for cluster-randomized trials provided in the Cochrane Handbook for Systematic Reviews (Higgins 2008a). If trialists have failed adequately to control for the clustering effect in their analyses, we will request individual participant data to calculate an estimate of the intra-cluster correlation coefficient (ICC). If these data are not available, or cannot be obtained, we will identify external estimates of the ICC from similar studies or other resources. Having established an appropriate ICC, we will reanalyze the data, which we will then enter into RevMan (RevMan 2014) to calculate effect sizes and confidence intervals using the generic inverse variance method (Higgins 2008b). If insufficient information is available to control for clustering in this way, we will enter data into RevMan using individuals as the unit of analyses, and perform sensitivity analyses to assess the potential bias that may have occurred as a result of the inadequately controlled clustered trials. We will also perform sensitivity analyses if the ICCs were obtained from external sources.

Cross-over trials

Cross-over designs are not anticipated in this review, but in the event that we find any, we will combine the results of the cross-over trials with the results of the parallel-group trials. If data from a cross-over trial are restricted or cannot be obtained from the authors, we will use the presented data within the first phase up to the point of cross-over. We will pool data from cross-over trials according to the methods described by Higgins and Elbourne and colleagues (Elbourne 2002; Higgins 2008b).

Studies with multiple treatment groups

If a study were to compare two types of reading interventions and a third group not receiving treatment, we will first confirm that each intervention would individually meet all the inclusion criteria. We will then determine whether the two intervention groups received equivalent or different treatments. If the groups received equivalent treatments, we will combine the data from the two treatment groups to form one pair-wise comparison with the control group. If the study contains groups receiving different interventions, we will form two pair-wise comparisons by splitting the sample size of the control group and comparing each intervention group separately with the reduced sample control.

Dealing with missing data

We will assess missing data and dropouts in the included studies. We will investigate and report reasons, numbers, and characteristics of dropouts. We will attempt to contact the authors of trials when further information or data are necessary. For studies in which the missing data are not available, we will conduct analyses using only the available data; i.e., we will not impute missing data. We will consider the extent to which missing data may impact the results of the review and assess the sensitivity of any primary meta-analysis to missing data using the strategy recommended by Higgins (Higgins 2008b).

Assessment of heterogeneity

We will first assess clinical variation across studies (e.g., participant factors, curricula used in trial, outcome assessment) and trial factors (e.g., randomization concealment, type of intervention, type of comparison group). We will then examine heterogeneity among included studies through the use of the Chi2 test, where a low P value indicates heterogeneity of treatment effects. We will also use two measures to assess the extent of heterogeneity across studies. The first will be the Tau2, which provides an estimate of the variance of the true effect and is represented at T2. The T2 metric quantifies the extent of heterogeneity between studies and will provide information on the dispersion of effects across included studies. We will also use the I2 statistic to determine the percentage of variability that is due to heterogeneity rather than sampling error or chance (Higgins 2002).

Assessment of reporting biases

If we identify 10 or more studies reporting the same outcome, we will draw funnel plots (estimated differences in treatment effects against their standard error). Asymmetry could be due to publication bias, but can reveal a real relation between trial and effect size, such as when larger trials have lower compliance and compliance is positively related to effect size (Sterne 2008). When such a relation is found, we will first examine clinical variation between the studies (Sterne 2008, 10.4). As a direct test for publication bias, we will conduct sensitivity analyses to compare the results from published data with data from other sources.

Data synthesis

The primary analyses will focus on the general effectiveness of reading intervention for children and adolescents with intellectual disability. We will synthesize studies including data for each relevant primary and secondary outcome in separate analyses for each type of intervention; we will only statistically synthesize studies using similar intervention types. We will synthesize the effect sizes drawn from the included studies that will represent intervention effects. These intervention effects could be indexed as either group differences in mean scores or differences in the odds of an event occurring (depending on how the outcome is measured). We will conduct the synthesis of these effect sizes using a random-effects model in which it is assumed that the variability across studies represents factors that are beyond subject-level sampling error alone. The rationale for the use of a random-effects model for the included studies is that there are likely to be several differences related to the samples (e.g., IQ level, etiology, age), settings (e.g., country effects, educational environment), and interventions (e.g., duration, length, intensity) that may not be readily reported in the studies. As such, we have deemed a random-effects model most appropriate for the present analyses. We will carry out these analyses for studies in which the samples, interventions, and outcomes meet our predesignated inclusion criteria.

Subgroup analysis and investigation of heterogeneity

We will explore moderator effects for chronological age, level of severity of intellectual disability, language skills, and intervention density (e.g., session duration, session frequency, intervention duration). We may conduct further investigation of the causes of heterogeneity using subgroup analyses. Possible subgroups that we may examine, if present, are: intervention setting and type of trial (e.g., parallel or cross-over).

Sensitivity analysis

In order to explore the impact of studies with high risk of bias on the robustness of the results of the review, we will conduct sensitivity analyses by removing studies with a high risk of bias (e.g., sequence generation, incomplete outcome data, blinding of outcome assessment) and reanalyzing the remaining studies to determine whether these factors affect the results. We will also reanalyze the data using different statistical approaches (e.g., using a fixed-effect instead of a random-effects model) to explore the impact of our choice of effect model (Higgins 2008a). If RCTs and quasi-RCTs are included together in a synthesis, we will explore the impact through sensitivity analyses by removing the quasi-RCT studies. Moreover, we will examine the influence of different procedural decisions taken by the review authors (e.g., use of different dependent variables, exclusion of outlying studies) through sensitivity analyses to determine the impact of the decisions on the overall results.

Acknowledgements

This protocol was produced within the Cochrane Developmental, Psychosocial and Learning Problems Group.

Contributions of authors

BR, CL, DM, and DH contributed to the development of this protocol. CL and DH drafted the introduction and BR, CL, and DM drafted the methods. BR, CL, DM, and DH reviewed and approved the full protocol. BR and DM will screen the abstracts and titles, retrieve potentially eligible papers, and make decisions about eligibility. CL and DH will independently extract data. We will resolve any disagreements by conference until there is agreement or by mediation with a third party (CL for study selection and BR for data extraction). BR, CL, DM, and DH will draft and approve the full review.

Declarations of interest

Brian Reichow - receives royalties for the publications Evidence-Based Practices and Treatments for Children with Autism, and Adolescents and Adults with Autism Spectrum Disorders. He receives honoraria from lectures on developmental disorders, salary support from the University of Connecticut Health Center, and funding from the US Department of Education and the US Health Resource Services Administration Bureau of Maternal and Child Health, none of which supported or influenced his work on this protocol.

Christopher Lemons - receives honoraria from lectures on disabilities and education, former salary support from the University of Pittsburg, current salary support from Peabody College of Vanderbilt University, and funding from the US Department of Education, none of which supported or influenced his work on this protocol.

Daniel Maggin - receives honoraria from lectures on disabilities and education, salary support from the University of Illinois at Chicago, funding from the US Department of Education and consultancy fees from the National Center for Intensive Interventions, serving as consultant on behavioral interventions and progress monitoring, none of which supported or influenced his work on this protocol.

David Hill - none known.

Ancillary