Types of studies
We intended to include randomised controlled trials (RCTs) and quasi-RCTs in this review. We considered within-subject studies, in which the order of presentation of devices was randomised, as quasi-RCTs. Within-subject studies are similar in design to conventional cross-over studies, but instead of offering interventions sequentially, low vision aid (LVA) studies frequently offer several types of aids sequentially in one study session and also measure outcomes sequentially, in the same session.
Types of participants
We intended to include trials involving children between the ages of 5 and 16 years with low vision as defined by, or equivalent to, the WHO 1992 definition (World Health Organization 1992): "A person with low vision is one who has impairment of visual functioning even after treatment and/or standard refractive correction, and has a visual acuity of less than 6/18 to light perception, or a visual field of less than 10° from the point of fixation, but who uses, or is potentially able to use, vision for the planning and/or execution of a task". In logMAR equivalents, this may equate to visual acuity worse than 0.48, but better than or equal to 2.7 logMAR (Schulze-Bonsel 2006).
We decided to exclude pre-school age children, as young children tend to hold objects close to their face to achieve magnification and LVAs are not usually prescribed to this age group. If LVAs, including electronic aids, are presented to children under the age of five years, the aim is to introduce children to the concept of electronic devices in a playful manner and not actually to improve access to visual information.
Types of interventions
We planned to include studies that explore the use of assistive technologies (ATs). These would include all types of closed circuit television/electronic vision enhancement systems (CCTV/EVES) and computer technology including tablet computers and adaptive technologies such as screen readers, screen magnification and optical character recognition (OCR). We intended to compare the use of ATs with standard optical aids, which include distance refractive correction (with appropriate near addition for aphakic (no lens)/pseudophakic (with lens implant) patients) and monocular/binoculars for distance and brightfield magnifiers for near. We also planned to include studies that compare different types of ATs with each other, without or in addition to conventional optical aids, and those that compare ATs given with or without instructions for use.
A separate Cochrane review has explored the effects of optical aids in the same population (Barker 2014). The motivation to split the topic into two reviews lies in the difference in what these technologies try to achieve. Optical aids are prescribed to facilitate reading and access to printed material by providing magnification. Some electronic ATs have a broader aim: facilitating access to education, but also to social media and real-time information available via the internet, for example maps/directions, educational or leisure activities offered in the vicinity etc. As such, a comparison of optical aids with ATs has to be limited to outcomes on which both types of devices can have an effect, such as reading and access to educational materials.
Types of outcome measures
Low vision affects many aspects of a person's life. Interventions aim to improve one or more different area(s) of difficulty. Outcome areas relevant to low vision include mobility, activities of daily living (ADL), self esteem (happiness, mental health), literacy (reading, writing, access to information), visual functioning, use of LVAs, social contact/participation, use of technology and employment (Douglas 2013). A recent systematic review of the effectiveness of low vision service provision categorised outcomes into five groups: objective/clinical outcomes, ADL/functional outcomes, vision-related quality of life (VRQoL), psychological status and general health-related quality of life (HRQoL) (Binns 2012). Outcome measures for objective outcomes include near visual acuity (VA), distance VA and reading accuracy, comprehension and speed (Binns 2012). A range of questionnaires is available to measure functional outcomes relating to ADL, psychological status, VRQoL and HRQoL, such as the Manchester Low Vision Questionnaire (MLVQ) (Harper 1999), the Low Vision Quality of Life Questionnaire (LVQoL) (Wolffsohn 2000), the National Eye Institute Visual Function Questionnaire (NEI-VFQ) (Mangione 1998; Mangione 2001), and the Impact of Vision Impairment Profile (IVI) (Hassell 2000; Weih 2002). Only a few tools have been developed and validated for use in children and young people, and even fewer have been developed with focus groups of children and young people. Examples include the Impact of Vision Impairment Profile for Children (IVI_C) (Cochrane 2011), the Cardiff Visual Ability Questionnaire for Children (CVAQL) to assess VRQoL (Khadka 2010), the Functional Vision Questionnaire for Children and Young People with Visual Impairment (FVQ CYP) (Tadić 2013), and the general health-related Pediatric Quality of Life Inventory (Varni 2001; Varni 2002).
The aim of ATs is not to improve all of the above outcomes. Rather they aim to assist with specific visual tasks. In an educational setting they are intended to improve access to written material from the conventional curriculum. They may also be useful for independence in ADL. Outcome measures potentially appropriate to evaluate the effectiveness of LVA ATs are therefore those related to vision-related quality of life, as well as measures of visual function related to reading (for example, reading speed) and literacy (reading accuracy and comprehension).
Reading performance has been found to be one of the best predictors of patient-reported visual ability and VRQoL (Hazel 2000; McClure 2000). Reading is an important function in daily life. It is a standard outcome in studies monitoring conditions causing visual impairment and in clinical trials evaluating the effectiveness of interventions (Rubin 2013). Reading speed may be the most appropriate primary reading-related outcome, as it evaluates the functional visual effect of the aid. CCTV, electronic reading aids, tablet computers and mobile phones can all be used to scan and enlarge text. Maximum reading speed may be the most commonly used outcome in assessing the effect of reading aids, and is the primary outcome explored in a Cochrane review on reading aids for adults with low vision (Virgili 2013). It is typically stable across a range of print sizes over a certain threshold (critical print/font size), whereas at smaller print sizes, below the critical print/font size, the reading speed slows and the reading acuity limit is reached (Ahn 1995a; Ahn 1995b; Bailey 2003). Using standardised reading charts such as those in the Minnesota Low-Vision Reading test (MNREAD), a plot of reading speed against font size (adjusted for reading distance and expressed in logMAR) can be obtained (Legge 2007). Typically, reading speed also slows above a certain magnification due to the restricted field of view and a lack of a proportional increase in the size of saccades (fast movements of the eyes) (Dickinson 2000).
The use of different font sizes in various studies is a methodological problem for meta-analysis. The most recent update of the Cochrane review on reading aids for adults with low vision included only studies assessing reading speed "when reading ordinary print size", i.e. 10 to 14 points (Virgili 2013). However, there is no universal agreement on ordinary print size for children. Books for young readers frequently use a large font size, i.e. 14 points or larger. School textbooks frequently reduce font size as their target audience matures, but there are no standards, and no recommendations as to when 'standard adult font size' (usually 9 to 14 points) should be used.
The type of reading material also influences reading speed. Research studies often use standardised reading charts such as the MNREAD and, more recently, the International Reading Speed Texts (IReST). Repeated, standardised assessment of reading performance requires a collection of texts of similar difficulty. Whilst the MNREAD chart contains single short sentences, IReST consists of 10 paragraphs of text (around 130 words each) and offers the advantage of a longer paragraph, which facilitates more accurate measurement of reading speed and judgement of fluency and mistakes (Trauzettel-Klosinski 2012). IReST has been evaluated in a cohort of normal sighted young adults and in patients with age-related macular degeneration, but has not been validated in children and young people.
In addition to reading performance, literacy outcomes, such as reading accuracy and comprehension, can give additional functional information. A measure of reading ability used in children with vision impairment is the Neale Analysis of Reading Ability (NARA), currently available in its second edition (NARA II) (Neale 1997). This is a comprehensive assessment of reading ability aimed for use with pupils aged 6 to 12 years, and is also recommended for use beyond the age of 12 years in children with sensory impairment. The test material consists of six paragraphs that increase in length from 26 to 140 words, and increase in difficulty. The test is designed to assess oral reading ability in terms of reading rate, accuracy and comprehension. Validation data are available for normally sighted individuals, and also for children and young people with visual impairment (Douglas 2002; Hill 2005). There are two parallel versions of the test, which permits the same child to be re-tested without remembering a previous test and thereby altering the score. The child's scores are converted into reading ages for accuracy, comprehension and speed. Accuracy is determined by noting reading errors such as mispronunciations, substitutions, refusals, additions, omissions and reversals. Comprehension is measured by asking the child a number of set questions concerning the passage he or she has just read. Reading speed is measured by timing the passages read and converting this into words per minute over the total number of passages read. Results can be plotted as graphs comparing the performance of VI students with normal-sighted age-matched peers (Douglas 2002; Hill 2005).
All literacy evaluations need to take into account that children are learning to read, i.e. are developing a skill. Children with low vision often read print more slowly and less accurately than normal, sighted peers (Douglas 2004; Gompel 2004). Comprehension may also be delayed; this may be linked to general delay in reading development (Douglas 2002). Other literacy tests used in educational settings, such as the National Foundation for Educational Research (NFER) and Access Reading Test (ART), include access features for children with low vision (enlarged print, braille, extended time), but no data from children with low vision are available.
Usage of ATs is a further important primary outcome measure. ATs are more costly than optical and non-optical aids. Peer pressure and the fear of 'standing out' may lead to optical aids being used infrequently or abandoned (Mason 1999). However, as the use of technology is mainstream, the acceptance of technological solutions even with adaptive technologies may be higher than with conventional optical aids. Electronic ATs are regularly provided to children of school age by the Educational Authority to tackle specific functional tasks within the classroom environment. The use of these ATs is limited by the acceptance of these devices by the child, in addition to other practical implementation factors such as the training and support of teachers and support staff and the day to day issues of moving equipment to different locations and equipment maintenance.
Lastly, with a view to costs of purchase and maintenance, the useful lifespan of a device is a relevant point. The lifespan of AT may be longer or shorter than that of conventional optical aids; due to their different capabilities, including magnification functions, one device may also be useful to the same user for longer despite potential worsening or improvement in visual function.
The following outcomes will have been assessed using a standardised chart such as MNREAD or IReST, or a standardised literacy test such as NARA.
Reading accuracy as errors per words read.
Reading comprehension as number of correctly answered set questions concerning the text read.
Reading acuity in logMAR, defined as the smallest print/font that the child/young person can read without making significant errors.
Critical print/font size, defined as the smallest print/font that the child/young person can read with maximum speed.
Fatigue-free reading duration in minutes.
The following outcomes have been measured with a different means of assessment (i.e. not standardised chart or literacy test).
Acceptance of the LVA, as reflected in usage (days per week, hours per day, at home and at school).
Independent learning, i.e. ability to access the curriculum independently, as assessed by questionnaires.
VRQoL, evaluated using any validated VRQoL scale for children.
HRQoL evaluated using any validated HRQoL scale for children.
Useful lifespan of device.
With regard to the time points of evaluation, general child development and, particularly, the development of reading and literacy skills will affect the effect size of interventions at given time points. One would expect an increase in reading speed with time as a younger child learns to read, regardless of LVA use, but using an aid may allow faster development of reading skills. On the other hand, a child's ability may have improved to a degree over that period of time, just as his/her general development has progressed.
For this review, we intended to consider the following time points:
Primary outcome: 3 and 12 months (+/- 3 months) after the intervention and relevant instructions, if any, have been issued, where three months is a proof of concept.
Secondary outcomes: 12 months (+/- 3 months). Useful lifespan of device may exceed the second time point; we will note data if available.
Ultimate outcomes such as educational attainment, as measured in educational progress, would be desirable, but due to the length of follow-up required, these are unlikely to be captured in research studies.
No adverse outcome is expected, and any adverse effects on visual function would be detected by primary and secondary outcome measures. We planned to summarise narratively any unexpected adverse outcome reported by study authors.
No systematic review of economic data has been conducted in this review, but we intended to report on the unit cost of devices as well as on costs of healthcare personnel involved and overall health service cost for each programme when available in included studies.