7. Vision screening before age four years
Chapter editors: Anna Horwood, Maria Fronius
a. General background
Identifying and treating reduced visual acuity (VA) in early childhood is the main purpose of screening. Testing VA using logarithmic tests cannot be done accurately in very young children due to their lack of cooperation and cognitive immaturity. Early childhood vision screening is broadly divided into two areas:
- Firstly, neonatal vision screening which aims to detect major ocular pathology and risk of severe vision loss in the first weeks of life. It has very different objectives from vision screening beyond early infancy and does not target the reduced visual acuity and[popup_anything id=”3318″]that is the target of later screening. With the exception of section c, the majority of this chapter is devoted to neonatal screening.
- Beyond the neonatal period, many children under 4 years of age will be unable to do a linear logarithmic VA test reliably, so between infancy and 3-4 years amblyopia may be suspected or detected, but only poorly quantified. There is significant controversy about whether, when and how, children should be screened (see section c ‘vision screening from infancy up to 4 years’ below and chapter 8 on photoscreening). Section c outlines the issues, but the consensus from the EUSCREEN study is that by introducing screening before accurate VA testing is possible, costs are increased for only modest improvements in outcome.
b. Neonatal vision screening
The main purpose of vision screening in the[popup_anything id=”3542″]is to detect ocular pathology and risk of severe vision loss, not the reduced visual acuity and[popup_anything id=”3318″]that is the target of screening at a later age. A large amount of visual development occurs in the first six months of life, so testing in the neonatal period, whilst essential, is a very poor predictor of later visual problems such as [popup_anything id=”3371″]or amblyopia1 2.
Infants are born with very limited [popup_anything id=”3353″](VA), poor ability to detect contrast, a wide range of refractive errors, inactive focusing (accommodation), immature binocular vision, unstable eye alignment. They are attracted to faces, lights/ windows and high contrast images, and may be very slow to change fixation to new targets (‘sticky fixation’). During typical development different visual processes have different developmental trajectories:
- Visual acuity improves dramatically due to neural and ocular growth from logMAR 1.5 (6/180) or worse at birth to logMAR 0.4 (6/15) at 12 months and logMAR 0.2 (6/9.5) or better at five years3 (for details of notation, see Appendix 1).
- A wide range of refractive errors found at birth grow towards normal (emmetropisation). Refractive errors outside a broad envelope may not emmetropise, and those beyond certain age-related limits may persist or develop into later life. Emmetropisation occurs mostly in the first two years of life, then slows and is largely complete by teenage years. In some children the emmetropisation process is defective (for example in children with Down syndrome, prematurity or developmental delays) so their early refractive error does not normalise. It is not possible to predict with any certainty whether an infant with a refractive error will emmetropise or not.
- Myopia (short sightedness) is mostly absent in early childhood but develops and increases later. The earlier the onset of myopia, the more myopic the child is likely to be, increasing the risk of developing high myopia later in life, the highest risk factor for visual impairment second to age.
- Binocular vision is very rudimentary at birth and stereo vision is not present. “Adult-like” motor control and depth perception develop relatively suddenly and are relatively mature by four months of age, followed by much smaller changes in later infancy and childhood. If binocular vision does not develop normally, many children will develop[popup_anything id=”3477″]beyond the neonatal period, although it is not clear whether very early onset strabismus is a cause or consequence of lack of binocular vision development.
For these reasons, only screening for the most severe sight-threatening conditions is possible or indicated in early infancy (the first six months of life).
i. Pre-implementation considerations and preparations
Consideration of the political, cultural, geographic, economic and demographic elements in preparation for implementation are covered in chapter 4 of this manual. Neonatal screening for the most common sight-threatening conditions is common in most countries, so infrastructure considerations are mainly to do with tailoring existing services maximizing coverage, efficiency and reducing loss to follow up. If a completely new service is to be set up, primary considerations are listed below:
- whether the proposed programme is local, regional, or national, and how reporting will be fed upwards to national datasets.
- funding – both for initial implementation and for sustainability.
- what kind of maternity care expectant mothers receive e.g. will provision have to be made for babies who leave hospital soon after birth, or are born at home.
- how will babies needing special care be screened? Premature infants have additional visual risks (see next section) and need to be placed on alternative care pathways.
- will the visual screening be carried out by staff carrying out other health checks e.g. hips, hearing, or by those with specialist training.
- whether the necessary follow-up diagnostic testing facilities exist in terms of appropriate venues and equipment and if so, how these are accessed. Conditions such as congenital cataract and retinal tumours are rare and may be treated by tertiary centres requiring many hours travel, which may be impossible or unaffordable for some parents, so will provision be necessary to support them.
- will special efforts be necessary to ensure that parents understand the importance of the screen and, particularly, the urgency of rapid early diagnosis and treatment.
ii. Types of vision screening in the neonatal period
Newborn vision screening targets severe sight-threatening ocular diseases such as cataract, neonatal ocular infections, corneal opacities and ocular tumours. It is vital that these are detected very early because they either can be life-threatening, or lead to a very severe form of amblyopia (‘stimulus deprivation amblyopia’) which can be prevented by appropriate, and often intensive, treatment which must be started within the first weeks of life. These conditions are rarer than other forms of amblyopia (for example the prevalence of congenital cataract is less than 0.05%4 and of neonatal tumours is less than 0.00008%5 compared with amblyopia prevalence of around 3%6).
Screening of specific at-risk groups
Some children are at more risk of poor vision than others. Children born prematurely or of low birth weight are specifically at risk of retinopathy of prematurity (ROP) which occurs due to abnormal development of the retina and its blood supply associated with the pre-term delivery and neonatal intensive care. ROP is a leading cause of childhood blindness if left untreated. ROP screening in the neonatal period is offered to premature or very low birth weight infants only, and involves regular detailed retinal examinations after dilating eye drops, carried out by experienced paediatric ophthalmologists. It is therefore much more intensive and targeted than general neonatal screening and will not be covered in detail in this document.
Children at risk of metabolic, genetic or inherited diseases which affect the eyes, for example phenylketonuria, deaf children and children with disabilities, may also need specific testing throughout infancy. Refractive error does have a genetic component, but many early refractive errors will emmetropise, so neonatal screening for refractive error is not indicated.
iii. Objective setting
Neonatal screening is common in many countries but the timing of these screenings varies (as evidenced by the EUSCREEN Country Reports). Neonatal vision screening can take place in maternity units or in the community, and may be repeated in the first weeks of life to ensure that difficult-to-test infants are tested properly and that emergent conditions which may be minimal in the immediate neonatal days, such as infections, developing cataracts or haemangiomas are detected and referred.
Testing is commonly carried out by general medical or paediatric personnel such as paediatricians, neonatologists, GPs, nursing or midwifery staff in the course of more general health checks.
iv. Target conditions
For neonatal screening the target conditions are ocular media opacities (corneal opacities, cataract and ocular tumours), conditions which cover the pupil (lid ptosis, large lid haemangiomas) and signs of ocular infections (red or swollen eyes). Any of these can prevent clear images reaching the retina and prevents vision developing normally. Such abnormal visual experience is extremely harmful and needs prompt treatment. Some, such as retinal tumours, can be life-threatening.
The screening may take place in a maternity unit if the infant is in the unit for long enough for a reliable test to take place. In units where mothers are discharged soon after birth, or if birth takes place at home, it might be better for the test to take place during other health checks in the neonatal period in the community or on a home visit. As with all screening that depends on parents bringing their infants to be checked, uptake depends on parental awareness, acceptance, willingness and ability to attend. Where neonatal screening does not already exist, public information campaigns may be necessary to increase parental uptake of screening. Neonatal vision screening shares many similarities with neonatal hearing screening (see chapter 5). Coordination is necessary between maternity and community neonatal support networks.
vi. Information for parents
Neonatal vision screening is generally quick, and often carried out during other neonatal checks by nurses, midwives, paediatricians or GPs. The consent process is usually part of consent to general neonatal checks. Any opt-in system is likely to result in lower uptake of the screening from the most vulnerable groups. For more information see chapter 4e.
This basic screening is an opportunity to deliver basic eye care advice and alert parents to warning signs to watch out for in later infancy. These include:
- a white pupil
- strabismus emerging or worsening after three months of age. Occasional intermittent strabismus in the first weeks of life is common and should only be referred if still present at four months of age
- wobbling or unstable fixation (nystagmus or roving eye movements)
- anything that prevents either eye seeing e.g a persistently closed eye
- many children have watery and intermittently sticky eyes in infancy. Most will resolve spontaneously over the first year of life. Parents should be informed of how to seek advice for this or other issues
If a referral is made, parents must be made aware of the significance of the finding and that follow-up for diagnostic testing is important and urgent.
Unless parents are aware of some specific inherited condition, a family history of eye problems or glasses is frequently unreliable due to poor public awareness of different types of eye condition.
vii. Information for follow-up care providers
Ophthalmologists or others receiving referrals should be given clear details of each referral made: name, contact details, date of birth, reason for referral, any information they may need about triage of appointments (for example mild versus severe defect). Referral of conditions, such as congenital cataract or tumours may be to specialist centres out of area, so mechanisms should be in place to follow up referral outcomes.
viii. Screening personnel and training
Neonatal screening is frequently carried out by trained medical personnel during general neonatal checks. Using eye trained personnel may be less cost effective, especially in community settings or smaller units, when only a few children need screening at a time.
Training materials and instruction of these professionals should be overseen or delivered by ophthalmology services and clear referral procedures should be defined.
Record keeping and communication become more complex if multiple organisations and record systems are involved e.g. hospital and community services, than if health records are more integrated, so great attention should be paid to:
- Identifying children to be screened and any that might be missed e.g. home deliveries
- How many are screened from this population (coverage)
- Reporting outcomes services receiving referrals to parents
- Audit and monitoring systems (collect data and report on coverage, referrals, true positives, false positives)
- Follow up data from referrals so that monitoring of outcomes can take place
ix. Minimum theoretical/practical requirements after training
All screeners should have an understanding of the conditions that the screening targets, and have a basic knowledge of their significance and management after referral. They should be able to perform a basic external examination of the eyes, test the ability to fix and follow and to check for a normal red reflex in the pupil. The target conditions are rare, so once the techniques have been taught, training is likely to involve the use of photographs, online resources or model eyes.
High quality training and assessment is essential and should ensure that screeners:
- understand the principles of screening and the difference between screening and diagnostics
- recognise and understand the potential risks and harms, limitations and benefits of screening
- fully understand the workings of the chosen equipment, its use and care
- have extensive hands on experience with the equipment, a period of supervised practice and a competency check before independently screening babies
- are knowledgeable about the policies and procedures of the hospital or clinic in which they are screening and how the screening programme fits in to routine practice
- are knowledgeable about and confident with the screening protocols
- are knowledgeable about the full screening pathway
- understand and can accurately document results and use the data management system
- can communicate sensitively and clearly with parents
- understand about sight-threatening vision conditions, their impact on development, and the interventions which are available
- are confident and calm dealing with babies and new parents
- understand the importance of working as part of a team with both other members of the screening programme and other professionals
x. Alternative training plans
When planning a new screening programme, it may be considered preferable to train many screeners at once, which lends itself to training days and group teaching. Once a programme is established, regular update/refresher days need to be planned to maintain quality standards, team building and motivation. New staff may need to be trained individually, with or without online resources. It is important that if trained in the field by another screener, high standards of the trainer are monitored and assured and bad habits do not creep in. Screeners may be specialists or senior trainees in other medical fields e.g. neonatal medicine, with high staff turnover due to training rotations, so ongoing, high quality training of all new staff must be ensured.
xi. Resources for training materials
As well as face-to-face training, written and online learning materials should be provided and regularly updated. Every new screener should have a supervisor or mentor for day-to-day advice if necessary. Most screeners will work in isolation from other screeners, so regular opportunities to meet or share experiences are recommended. In remote areas, this might need to be online.
xii. Follow-up of screening personnel and training
As a screening programme becomes established, expertise and community acceptance become embedded. Regular evaluation will help determine the communication needs and the interval between training and re-training. Feedback to individual screeners about the diagnostic outcome of their referrals is important in terms of quality assurance and screener confidence.
xiii. Protocol: test choice
The tests generally include:
- external observation, looking for any corneal opacity, a white, atypical or obscured pupil, abnormal iris pigmentation or anomalies such as coloboma or aniridia, nystagmus, abnormal redness, sticky eyes, lid abnormalities, albinism
- ability of the infant to fix and follow the examiner’s or the mother’s face as it moves slowly within the central visual field. Many newborns prefer to look at bright lights or windows in preference to a stranger’s face and will maintain fixation as their body is gently rotated, so this is also acceptable
- ‘red reflex’ testing. When looking through any ophthalmoscope set at zero a red reflex from the retina will be seen in the pupil. This demonstrates clear ocular media between the cornea and the retina (however, the red reflex test will not detect peripheral retinal abnormalities such as peripheral tumours). A corneal opacity, cataract or other media opacities such as vitreous anomalies or central ocular tumours will show as an absent, dark or dim reflex7. Central lens or corneal opacities show up as a black dot in the red reflex. It is very important that testers are familiar with the range of normal reflexes because eye and skin pigmentation can cause a normal reflex to be anywhere between bright red in very blond children to a very dull, dusky reflex in children with dark skin
- Brückner Testing. The Brückner Test is an extension of the red reflex test, and requires very little extra training8. Instead of using the ophthalmoscope to look at one eye, the tester also sits more distantly so that the ophthalmoscope beam straddles both eyes. The two eyes are more easily compared and if the reflex is brighter at the top or bottom of the pupil, and especially if this differs between the eyes, it may indicate a refractive error that might need more careful monitoring
xiv. Protocol: rescreening steps
Some infants will be asleep, crying or inattentive at the time of testing so may need to be tested at another visit. If infants are discharged from maternity units before the screen, systems must be in place to make sure they do not miss the screen.
xv. Protocol: pass/refer criteria
The infant should exhibit:
- Normal external appearance.
- Ability to fix a face or bright light as it moves slowly in the central visual field.
- Clear red retinal reflex in each eye, symmetrical between the eyes.
If these are not all demonstrable, re-testing or referral is indicated.
xvi. Protocol: follow-up
Referral of children suspected of a defect should be to local ophthalmology services and should be urgent. A decision must be made whether a direct referral is made the service, or whether parents are expected to seek care themselves (which risks higher loss to follow up and should be avoided if at all possible).
Mechanisms should be in place that referrals are followed up. Ophthalmologist involvement is vital to successful evaluation, so referral and feedback mechanisms should be as minimal and efficient as possible to prevent loss of data. For example an existing database can be used for data reporting and referral, or a simple return postcard provided. In regions where private providers offer care, loss to follow-up and feedback issues need to be considered carefully. Tracing outcomes from screening is more difficult if community /specialist communication is not routine, healthcare data is not centrally shared, or if parents are left to make their own diagnostic appointments.
xvii. Communicating results to parents
The importance of the referral should be made clear to the parents so that diagnosis and prompt treatment can be as soon as possible. Parents must understand that delay in referral can be sight-, or even life-, threatening and must not wait until the child is older. As with neonatal hearing screening, this should be handled sensitively, with a mechanism for parents to access support and advice beyond the screening event.
Efficient monitoring of a screening programme is necessary to be able to carry out effective quality assurance, evaluation and reporting. Regular, appropriately-funded evaluation should take place, including assessment of coverage, training of screeners, performance of screeners, method of screening, referral criteria and results. More detailed information on monitoring can be found in chapter 11.
Note that any data registry should comply with applicable legislation (see chapter 4e) and that, for newborn screening, quality evaluation is especially difficult because referrals are rare and many professionals will make very few, if any, referrals.
xix. Adapting an existing programme
Most countries have some type of neonatal vision screening, but adaptations and efficiencies may still be possible or necessary. For example, changes in when or where the test takes place, and by whom. Before change is implemented, evaluation of current services must take place, so that any effects of the change can be properly monitored.
xx. Overcoming barriers
Barriers to setting up or improving any screening programme should be identified at the earliest stage, in relation to local circumstances. Identification and strategies to overcome them are critical to success. They may be very low level (local communication or transport issues), mid-level (training or quality assurance) or high-level (strategic or funding). A careful risk register of all possible barriers and how they will be handled and overcome should be kept and regularly updated.
c. Vision screening from infancy up to 4 years
The EUSCREEN Country Reports show that many countries screen children’s vision between 6 months and 4 years of age, but there is a wide variation in practice. Some children are not tested at all during this period, and referrals are made only in the presence of signs or symptoms causing parental or professional concern; while others are screened annually, using a wide range of tests and test combinations. Most assessments rely on orthoptic tests, such as cover tests, corneal reflection assessment, ocular motility testing, stereotests, prism tests and gross VA assessment; looking for strabismus and any clinical sign of reduced acuity in one or both eyes. Stereotests administered as a single screening test, on the premise that amblyopia and strabismus result in poor stereovision, have also been advocated9, but poor sensitivity means they have not been widely adopted. More recently, photoscreening or autorefraction looking for refractive risk factors for amblyopia are being added to some of these assessments (see chapter 8). All these tests generally target risk factors for amblyopia, rather than amblyopia itself. Referral rates of 53% were reported in a study using a battery of VA, binocular vision and photoscreening tests in 4-year-olds10. There is often very poor reporting and availability of outcome data, so comparisons are difficult or impossible.
There are specific issues relating to testing young children that make such screening of limited value in terms of most of the WHO recommendations for screening (see Chapter 2a).
- VA is often inaccurate, unreliable and poorly quantifiable because young children cannot do linear logarithmic tests. Young children are often inattentive, uncooperative or will not tolerate uniocular testing.
- Tests may be timed to coincide with vaccinations, hearing or other health tests, so children may be tired or apprehensive on the day.
- Testing of young children is a highly skilled process, and results from even the most skilled testers are more variable. VA assessment is often carried out by clinic nurses, health visitors, GPs or paediatricians with restricted eye testing expertise and who may use poor technique11.
- Because these tests need specialist skill to interpret and children can be difficult to test, sensitivity, specificity and[popup_anything id=”3349″]for amblyopia are low, and a high proportion of children may be referred for further assessment, while milder amblyopia may be missed.
- Tests may be timed to coincide with vaccinations, hearing or other health tests, so children may be tired or apprehensive on the day.
- Early refractive errors can change rapidly, or normalise, because[popup_anything id=”3534″]is active, especially before 2 years of age.
- Some conditions, such as accommodative strabismus, amblyopia and[popup_anything id=”3323″]develop during this period, so test results can change quickly. A pass one day may be a fail a month later.
- Children are not often in compulsory nursery or education, so parents must bring children to be tested, resulting in low or poor[popup_anything id=”3463″] particularly of the most at-risk children in disadvantaged groups, or where public health awareness is poor.
- More screening events cost more, but once children are referred, costs rise even further. The patient journey to discharge after amblyopia treatment is longer, because children are rarely discharged until after the most active phases of the critical period.
- Cost effectiveness calculations up to the point of diagnosis for these early and multiple screenings may be possible in areas where there is good community data, but often the costs lie beyond screening, borne by state health services or parents for much longer. Data sharing between primary and secondary care may be patchy or difficult.
- Audit is very difficult because even if a final VA after treatment is testable, a comparable test result at referral was not possible. A child may be known to have been amblyopic on referral, but precisely how amblyopic they were, or how much they have improved, is often unknown.
- Many children with strabismic amblyopia present due to parental concern about an obvious strabismus, not via screening, so many screened children may already be under secondary care on the screening visit. However the other main causes of amblyopia such as refractive error, may only be detected by VA or refraction, and are not detected by orthoptic tests.
The[popup_anything id=”3354″]Screening programmes: a short guide outlines the general issues very well. Decision-makers must decide whether multiple early and imprecise tests provide more benefit than reducing screening interventions to later, more accurate tests from the age of four years. The advent of photoscreening, possible in very young children (see chapter 8) has highlighted these controversies further.Although amblyopia outcomes are better if treatment is started early, differences are small and outcomes are still generally good even if treated later, as long as treatment is carried out within the critical period. For example, the large UK ALSPAC cohort study compared children screened 6 times up to 37 months, with a group tested just once at 37months. Amblyopia treatment outcomes were good in both groups (both better than a mean of 0.2 logMAR) and the intensively screened group only saw an average of 3 letters better than the later, single screened group)12 13 found little effect from removing a screening at 6-9 months, and modelled14 that omitting a further screening at 24 months would also not lead to significantly worse outcomes. Moving screening from 3-4 years including orthoptic testing, to a VA test alone at school entry 4-5 years in the UK made little difference to outcomes, but because parents did not have to bring their children to be tested, population coverage increased dramatically15. This clear advantage may not apply in countries with significantly later school entry age unless most children attend nursery or kindergarten before formal schooling.
Looking for strabismus, refractive error and other risk factors for low vision in very young children may change what actually becomes the target condition. Although a screening service may say it targets amblyopia and significant low vision, by referring children who are at risk for low vision, rather than children who actually have low vision, many more children will be referred and then receive treatment for a non-amblyopic condition.
Think box: which choice would you make?
Refer when low VA can be proved from a single VA test (4-5years) – e.g. the UK model:
– Only children with actual low vision receive treatment
– Amblyopia and everyday low vision remain the target conditions
– Mild, non-amblyogenic conditions will not be referred and so only be treated as they present later
– Low cost – fewer referrals, high[popup_anything id=”3349″]for the target condition
– Treatment for genuinely amblyopic children may not start until later in the critical period, so outcomes may be marginally worse, but are marginally worse outcomes significant to an individual’s functional or quality of life, or at a population level?
– Risk of harm rests on these marginally worse outcomes and possible adverse effects of mild conditions
Refer earlier from risk factor screening (inaccurate VA tests / orthoptic tests / photoscreening at under 4 years) (common in many high-income countries):
– Children who might have low vision receive treatment to mitigate or prevent amblyopia
– Risk factors become the target condition for the screening
– Marginally better amblyopia treatment outcomes
– Conditions other than low vision and amblyopia will be referred and treated (mild refractive errors, non-amblyopic, cosmetically insignificant strabismus, convergence or stereovision defects
– Evidence that these conditions are socially significant is still equivocal
– Once referred, amblyopic children need more appointments from expensive services and longer treatment and supervision
– Much more expensive – to parents and health services. Are your country and population able and willing to pay for it?
– Risk of harm rests on costs and potential for over-treatment of insignificant conditions.
- Teller DY, Movshon JA (1986): Visual development. Vision Res 26(9):1483-506.
- Atkinson J (2000): The developing visual brain. Oxford: Oxford University Press.
- Chandna A (1991): Natural history of the development of visual acuity in infants. Eye 5(1):20-6.
- Wu X, Long E, Lin H, Liu Y (2016): Prevalence and epidemiological characteristics of congenital cataract: a systematic review and meta-analysis. Sci Rep 6:28564.
- Moore SW, Satgé D, Sasco AJ, Zimmermann A, Plaschkes J (2003): The epidemiology of neonatal tumours. Report of an international working group. Pediatr Surg Int 19(7):509-19.
- Webber AL & Wood J (2005): Amblyopia: prevalence, natural history, functional effects and treatment. Clin Exp Optom 88(6):365-75.
- Subhi Y, Schmidt DC, Al-Bakri M, Bach-Holm D, Kessel L (2021): Diagnostic Test Accuracy of the Red Reflex Test for Ocular Pathology in Infants: A Meta-analysis. JAMA Ophthalmol 139(1):33–40.
- LaMattina KC, Vagge A, Nelson LB (2019): Can the Red Reflex Test Detect Unequal Refractive Error? J Pediatr 214:175-177.
- Lang JI, Lang TJ (1988): Eye Screening with the Lang Stereotest. Am Orthopt J 38(1): 48-50.
- Nishimura M, Wong A, Dimaras H, Maurer D (2020): Feasibility of a school-based vision screening program to detect undiagnosed visual problems in kindergarten children in Ontario. CMAJ 192(29):E822-E831.
- Sloot F, Sami A, Karaman H, Gutter M, Benjamins J, Sjoerdsma T, Simonsz HJ (2017): Semistructured Observation of Population-based Eye Screening in The Netherlands. Strabismus 25(4):214-221.
- Harrad RA, Williams C, Sparrow JM, Northstone K, Harvey I, ALSPAC Study Team (2002): Visual Acuity at 7 Years After Orthoptic Screening at Different Ages – Results of a Randomised Controlled Trial. Invest Ophthalmol Vis Sci 43(13):2941.
- Sloot F, Sami A, Benjamins J et al. (2014): Effect of omission of population-based eye screening at age 6–9 months in the Netherlands. Acta Ophthalmol 93(4):318-321.
- Sloot F, Heijnsdijk E, Groenewoud JH, Goudsmit F, Steyerberg EW, de Koning HJ, Simonsz HJ (2017): The effect of omitting an early population-based vision screen in the Netherlands: A micro-simulation model approach. J Med Screen 24(3):120-126.
- Carlton J, Griffiths H, Mazzone P (2019): BIOS VISION SCREENING AUDIT: Academic Year 2017-2018. Sheffield: The University of Sheffield.