Costs and effectiveness of two models of school-entry visual acuity screening in the UK

Abstract

Cost effectiveness of different visual screening modalities cannot be calculated without long-term outcome data. This paper reports detailed outcomes from a gold-standard UK recommended orthoptist-delivered screening (ODS) at 4–5 years in school, compared to a neighboring school-nurse delivered screening (SNDS), both feeding into the same treatment pathway. The target condition was reduced visual acuity (VA) of worse than logMAR 0.2 in either eye.
Available records from screening databases and hospital records were analyzed, comparing the two services wherever possible.
More screening data was available from the ODS. ODS: 5706 screened, 3.5% referred. False positives 6.5%, PPV 91.4%, sensitivity 97.9%, and specificity 99.8% for reduced VA. Cost per child with reduced vision detected £195.22, and per amblyope detected £683.28. The mean treatment cost per child with reduced VA was £331.68 and for amblyopia treatment was £458.65.
SNDS: 5630 screened and 3.8% referred (plus some referrals to local optometrists lost to follow up). False positives 34%, PPV 53.2%, sensitivity and specificity estimated as 89.3% and 98.67%. Costs to secondary services of false positives were seven times greater. The cost per child with confirmed reduced vision seen at the hospital was 46% more; and per amblyope detected was 39% more.
Outcomes for treatment post referral in both groups were similar and excellent. 86% of genuine referrals improved to within normal limits with glasses alone. Of 221 genuine referrals with final outcome data, all now have better than 0.2logMAR acuity in the better eye and only two (0.9%) have residual amblyopia in one eye worse than 0.4logMAR.
About 14–18% of children with reduced VA would have passed AAPOS photoscreening referral criteria.
An orthoptist-delivered single VA screen at 4–5 years is highly cost effective with good outcomes. The main contributing factors to success appear to be training and experience in accurate VA testing, the opportunity to rescreen equivocal results, and monitoring, audit, and feedback of outcomes.

Full text

Full Terms & Conditions of access and use can be found at
https://www.tandfonline.com/action/journalInformation?journalCode=istr20
Strabismus
ISSN: (Print) (Online) Journal homepage: https://www.tandfonline.com/loi/istr20
Costs and effectiveness of two models of school-
entry visual acuity screening in the UK
Anna Horwood, Deborah Lysons, Victoria Sandford & Greg Richardson
To cite this article: Anna Horwood, Deborah Lysons, Victoria Sandford & Greg Richardson
(2021): Costs and effectiveness of two models of school-entry visual acuity screening in the UK,
Strabismus, DOI: 10.1080/09273972.2021.1948074
To link to this article: https://doi.org/10.1080/09273972.2021.1948074
© 2021 The Author(s). Published with
license by Taylor & Francis Group, LLC.
View supplementary material
Published online: 05 Jul 2021.
Submit your article to this journal
View related articles
View Crossmark data

Costs and eectiveness of two models of school-entry visual acuity screening in
the UK
Anna Horwood
a,b
, Deborah Lysons
b
, Victoria Sandford
b
, and Greg Richardson
b
a
University of Reading, Earley Gate, Reading;
b
Royal Berkshire Hospital, Orthoptic Department, Royal Berkshire Hospital, Reading
ABSTRACT
Cost eectiveness of dierent visual screening modalities cannot be calculated without long-term
outcome data. This paper reports detailed outcomes from a gold-standard UK recommended
orthoptist-delivered screening (ODS) at 4–5 years in school, compared to a neighboring school-
nurse delivered screening (SNDS), both feeding into the same treatment pathway. The target
condition was reduced visual acuity (VA) of worse than logMAR 0.2 in either eye.
Available records from screening databases and hospital records were analyzed, comparing the
two services wherever possible.
More screening data was available from the ODS. ODS: 5706 screened, 3.5% referred. False
positives 6.5%, PPV 91.4%, sensitivity 97.9%, and specicity 99.8% for reduced VA. Cost per child
with reduced vision detected £195.22, and per amblyope detected £683.28. The mean treatment
cost per child with reduced VA was £331.68 and for amblyopia treatment was £458.65.
SNDS: 5630 screened and 3.8% referred (plus some referrals to local optometrists lost to follow
up). False positives 34%, PPV 53.2%, sensitivity and specicity estimated as 89.3% and 98.67%. Costs
to secondary services of false positives were seven times greater. The cost per child with conrmed
reduced vision seen at the hospital was 46% more; and per amblyope detected was 39% more.
Outcomes for treatment post referral in both groups were similar and excellent. 86% of genuine
referrals improved to within normal limits with glasses alone. Of 221 genuine referrals with nal
outcome data, all now have better than 0.2logMAR acuity in the better eye and only two (0.9%)
have residual amblyopia in one eye worse than 0.4logMAR.
About 14–18% of children with reduced VA would have passed AAPOS photoscreening referral
criteria.
An orthoptist-delivered single VA screen at 4–5 years is highly cost eective with good out-
comes. The main contributing factors to success appear to be training and experience in accurate
VA testing, the opportunity to rescreen equivocal results, and monitoring, audit, and feedback of
outcomes.
KEYWORDS
Child vision screening;
orthoptist; amblyopia; cost;
effectiveness
Introduction
The United Kingdom National Screening
Committee (UKNSC) recommends that beyond
neonatal screening, the next child vision screening
should be a linear visual acuity (VA) test in the
first year of compulsory education at 4–5 years,
delivered or led by orthoptists.
1
The EUscreen study
2
is highlighting how cost
effectiveness modeling of different screening modal-
ities is severely hampered by lack of reports of long-
term costs and outcomes, in comparison to shorter-
term outcomes such as positive predictive values
(PPVs) for a specific diagnosis (for reviews see.
3,4
)
Even if data are audited locally, they are not shared.
Cost effectiveness is particularly important for
publicly funded health services paying for the
whole patient journey from detection to discharge.
Low screening costs may not be matched by equally
cost-effective follow-up, and outcomes may differ
between different screening timings and methods.
A particular issue in children’s vision screening is
comparative costs of early automated photoscreen-
ing for refractive risk factors for treatable reduced
vision and amblyopia, versus later visual acuity
(VA) testing by a skilled tester which detects actual
reduction in vision. Because more children will
have risk factors than will be amblyopic, low-cost
photoscreening
5
might be less cost effective in the
long term due to more, less precise referrals, and
CONTACT Anna Horwood a.m.horwood@reading.ac.uk Infant Vision Laboratory, School of Psychology & Clinical Language Sciences, University of
Reading, Earley Gate RG6 6ES, UK
Supplemental data for this article can be accessed on the publisher’s website.
STRABISMUS
https://doi.org/10.1080/09273972.2021.1948074
© 2021 The Author(s). Published with license by Taylor & Francis Group, LLC.
This is an Open Access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivatives License (http://creativecommons.org/licenses/by-nc-
nd/4.0/), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited, and is not altered, transformed, or built
upon in any way.

more expensive visits to secondary services, with-
out clear evidence of better outcomes at
a population level.
4
This paper describes a long-term audit of out-
comes and costs associated with referrals from two
established neighboring school-entry vision screen-
ing services in southern England from one
academic year, one following national guidelines
more closely than the other. Both populations are
majority White British, but with significant and
similar numbers of ethnic diversity including
South Asian, Afro-Caribbean and Eastern
European families (see Supplemental File) in both
areas. There are some areas of socio-economic
deprivation in both areas, but the whole region is
relatively economically prosperous.
Both services test 4–5-year-old children in their
first year of compulsory school, use the same vali-
dated linear logMAR test (the 3 m Sonksen test
6
with a pass threshold of 0.2logMAR in either eye).
They feed into the same healthcare provider man-
aged by the same National Health Service (NHS)
team, using common treatment protocols free at
point of delivery, and common records software.
All treatment of uncomplicated amblyopia and
refractive error is carried out by a joint orthoptist/
optometrist service managed by orthoptists. The
receiving hospital trust covers the whole county,
so few referrals go elsewhere, and local optometrists
generally refer any amblyopic child to hospital.
Orthoptist delivered screening (ODS)
This service is run by the orthoptic service and
delivered by five experienced orthoptists. Each
orthoptist travels to many schools in the area to
test large numbers of children (60–70 (two
classes) per session, >1000 per year), screening
once or twice a week, interleaved with hospital
practice. Children already in glasses are not
screened and do not feature in the analysis of
the data. Children are retested in school in the
next term if the test is a borderline fail (0.225–0.3
logMAR), or the orthoptist is not confident the
VA is accurate. As long as the child was
rescreened in the same academic year, their
data were included in the analysis. Audit and
feedback from the screenings are integrated
with hospital patient records, orthoptists can
access records of children they have screened,
and formal feedback from the service is shared
regularly.
School nurse delivered screening (SNDS)
This service is administered by the school nursing
service based in the local authority, not the health
authority, and feedback is patchier. The training is
not orthoptist led or delivered, as is recommended
by the UKNSC. Records held by the school nursing
service are difficult to access by the health services
due to data protection issues, and the health ser-
vices only feed back very basic data about referral
outcomes (and no treatment outcomes) to the
screening service.
The VA screen is part of other school entry
health checks, and each nurse is based in a small
number of schools, so they will test many fewer
children’s VA per session and per year. There is
no rescreen option, so all children not passing the
screen are referred.
Children with mild deficits of vision (0.3logMAR
in one eye and 0.2 in the other, 0.3 in both eyes, 0.4
in one eye, and 0.3 in the other) are referred to local
optometrists. For this study, we were unable to
establish how many children were referred down
this route, and this unknown number of children
are lost to follow up.
For further details of procedures and differences
between the services see the Supplemental File.
Methods
Outcomes from all referrals from children screened
in the academic year 2018–19 were scrutinized
from hospital records and data provided by the
screening services. As it was an audit, Ethics
Committee approval was not sought, but ethical
principles were followed, such as early anonymiza-
tion of identifiable data, and involvement of stake-
holders. Data from the ODS was much more
complete than the SNDS, both for structural rea-
sons and COVID-19-related difficulties in acces-
sing data.
Amblyopia was diagnosed on the first visit on
return 6–8 weeks after being prescribed any glasses
if the VA remained worse than 0.2 in one or both
eyes in the absence of pathology. To try to capture
2A. HORWOOD ET AL.

some idea of false negatives, which is not collected
systematically, all orthoptists in the county were
asked how many late-presenting children with
reduced VA they had seen from the cohort who
they considered to have been missed by the screen-
ing (not moved into the area later, presenting with
mild, non-amblyopic myopia which was likely to
have developed between the screening and referral,
or failed offered diagnostic appointments after
screening referral). The ODS service keeps the
screening records for audit purposes so it was pos-
sible to confirm if a child had passed the original
screen and orthoptists are encouraged to check.
Statistical analysis
Descriptive statistics were collected from available
records of children screened, referred, followed to
diagnosis, and followed to discharge. Where the
data were reliable, calculations or best estimates
were made for sensitivity, specificity, and positive
predictive values for the target condition of VA
worse than 0.2 in either eye.
Costs
Cost per screen included staff, administration, post-
age, screener travel, and any replacement equipment.
In this pre-COVID-19 period, there were few dis-
posables. Costs reported were based on median pay
points on national salary scales, travel at £0.25 per
mile, national postage rates, published NHS appoint-
ment unit costs,
7
payment made to local opticians to
supply spectacles at a national voucher rate, occlu-
sion patches, and manufacturer’s listed equipment
prices. At the time of analysis, the currency exchange
rate was £1 = €1.12 = 1.25 USD. Any additional cost
for the few children needing further pediatric
ophthalmology opinions were not included.
Results
Orthoptist-led service
5878 children entered school in the year (for details
see Table 1). 5839 (99.3%) were successfully
screened, including. 202 (3.5%) children who were
recalled for a second screening of equivocal results
before passing successfully. The remainder had
either special needs so were referred elsewhere
(10), were sick or from traveling families, so were
not in school on the testing day and were deferred
to the next year (25). Four were untraceable.
199 (3.5%) children were referred from screen-
ing. 48 (24.1%) of these were lost to first follow up
and a further 11 failed to attend during treatment,
leaving 140 (70.4%) children traced to discharge on
the audit date (July 2020).
Mean age at the new case visit was 5 years
0.3 months, with a mean time from referral to
reduced VA diagnosis of 42 days (range 8–122).
Mean age at audit was 6 years 3 months and most
had completed any active amblyopia treatment.
Fifteen children were still being monitored early
in 2020 when nonurgent appointments were can-
celed due to the COVID pandemic, so their final
outcome is slightly delayed.
Of the 151 children whose screening results were
followed up to diagnosis, 138 had confirmed
reduced vision (91.4% true positive for reduced
acuity). Thirteen (6.5%) were false positives, having
failed the screening but had good equal vision and
normal orthoptic assessment on their diagnostic
visit, and were discharged.
It is not possible to be certain of the number of
false negatives, but the poll of orthoptic department
staff suggests it is never more than three per year. If
this reasonable estimate is used, sensitivity for con-
firmed reduced VA and amblyopia of less than 0.2
logMAR in one or both eyes was 98%, specificity
was 99%. PPV was 91%.
129 children were refracted under cycloplegia
and 126 were given spectacles. Two subsequently
achieved normal VA without glasses, four were
diagnosed with pathology.
Twenty children (14.9% of children with con-
firmed reduced VA) had refractive errors which
would have passed the American Association of
Pediatric Ophthalmology & Strabismus (AAPOS)
refractive risk factor screening criteria (astigmatism
>1.5 D, anisometropia >1.5 D, and hyperopia >3.5
D)
8
if they had been photoscreened, but had sig-
nificant reduced vision or amblyopia.
On the follow-up visit 6–8 weeks later 74% of the
referrals reached better than 0.2 VA in each eye
with glasses alone. Thirty-six (26%) of the children
with confirmed reduced vision on their first visit
were still amblyopic, but 17 normalized after
STRABISMUS 3

a longer period of “refractive adaptation.”
9
So, of
the 138 children with confirmed reduced vision on
their first hospital visit, 79% improved with glasses
alone or spontaneously (no glasses indicated on
refraction).
Only 19 (13.7%) needed occlusion. Only one
child (with strabismic and anisometropic amblyo-
pia) had final VA worse than 0.4 (0.7 logMAR) and
eight others are still in the very final stages of
treatment with 0.3 VA or better. Four children
had subtle pathology on further investigation. The
remainder of the referrals are either stable with no,
or minimal, amblyopia and are discharged, or in
the final stages of observation before discharge to
their community optometrist.
The median number of visits for all children was
3 (range 1–9). Of children still amblyopic on return
with their first spectacles, the median number of
visits was 5 (range 3–9), although 8 are still under
supervision.
Costs (based on 2019–20 rates)
The total cost per screening test was £4.21 per child
(Table 2). The cost of a hospital appointment to
confirm reduced vision was £70 and the mean cost
of a glasses voucher for all the children prescribed
glasses was £46.59. The false positive screening
Table 1. Numbers and percentages of children targeted by the orthoptist-led (ODS) and School Nurse Led (SNDS) screening services. n/
a = data not available.
Orthoptist (ODS) School nurse (SNDS)
n % n %
SCREENED
Elligible 5878 n/a
Present at screening 5839 n/a
Parents declined (% of elligible) 4 (0.01) n/a
Already in gls so not screened (% of elligible) 133 (2.3) n/a
Absent on first screen (% of elligible) 256 (4.5) n/a
Recalled (% of elligible) 202 (3.5) not done
Special needs so referred elsewhere/sick /traveler/to be screened next year/
untraceable (% of elligible)
39 (0.7) n/a
Total screened 5706 5630
REFERRALS
Referred (% of screened) 199 (3.5) 215 (3.8)
Followed to diagnosis (% of referrals) 151 (75.9) 173 (80.5)
Traced to final outcome or audit date (% of referrals) 140 (70.4) 170 (79.1)
Observed further without refraction then discharged (% of referrals) 0 14 (6.5)
True +ve for reduced VA (% followed to diagnosis) 138 (91.3) 83 (38.6)
True -ve (estimated) 5504 (96.5) 5405 (96.0)
True +ve for amblyopia after first glasses 36 (26.1) 25 (30.1)
False +ve for reduced VA 13 (6.5) 90 (41.9)
False -ve (later presenting low VA) (estimated) 3 (10)?
Cycloplegic refraction 129 99
Spectacles prescribed 126 89
Parents declined treatment/went elsewhere 5 4
Good VA with glasses from optometrist on new case visit (discharged) 6 5
VA on screening below 0.3 both eyes (%of referrals) 29 (14.6) 29 (13.5)
(of the above not myopia or myopic (so likely poor Near VA too)) 1 10
Low VA under AAPOS refractive risk factor referral threshold (% of children with
low VA)
20 (14.5) 15 (18.1)
Age at follow up (years/months) 5 yrs 0mths 5 yrs 3mths
Age at full diagnosis (on return with glasses)(years/months) 5 yrs 2mths 5 yrs 5 mths
Mean delay from referral to diagnosis (weeks (range)) 6 weeks (1–17) 9 weeks (1–25)
Sensitivity for reduced VA 97.9% 95%CI 93.9%–99.6%
Specificity for reduced VA 99.8% 95%CI 99.6%–99.9%
Positive Predictive Value (PPV) for reduced VA 91.4% 95%CI 86.0%–94.8#% 53.2% 47.25%–59.07%
OUTCOMES
Followed to discharge or audit (% of referrals) 119 (59.8) 170 (79.1)
Pathology (% of referrals) 4 (2.0) 0 0
Needed occlusion after “refractive adaptation” (% of referrals) 19 (9.5) 11 (5.1)
Lost to full f/u (failed appointments or moved away after initial Dx) (% of genuine
referrals)
80 (57.9) 25 (30.1)
Final VA worse than 0.2 in the worse eye 8 (4.0) 6 All ongoing
Final VA worse than 0.4 in the worse eye 1 (0.7) 2 All ongoing
Final VA worse than 0.2 in the better eye 0 0 0 0
In final stages of treatment (% of genuine referrals) 15 (7.5) 18 (21.7)
Median number of visits (all children) 3 Range 1–9 3 Range 1–6
Median number of visits (amblyopes) 5 Range 3–9 5 Range 3–6
4A. HORWOOD ET AL.

referrals cost the NHS £910 in total. The cost to full
amblyopia diagnosis (screen + two hospital visits +
spectacles) was £189.80 per child. The cost per child
detected with confirmed reduced vision was
£195.22, and per amblyope detected was £683.28.
School nurse delivered service
Data collected, and available, was much less com-
prehensive, and over the COVID lockdown some
still is impossible to obtain (see Table 1).
About 5630 children were screened and 215
(3.8%) were referred, so referral rates were similar.
170 (79%) were followed to discharge or audit and
11 failed to attend during treatment. 20% were lost
to initial follow up. However only 83 referred chil-
dren had reduced vision confirmed at their new
case visit (true positive for reduced acuity 48%).
Seventy-three (34%) were clearly false positives
(VA 0.1 logMAR or better in either eye) and were
discharged after the first visit with normal VA
(compared to 9% for the ODS). A further 14 (8%)
had marginally reduced VA (0.125–0.2 logMAR in
one or both eyes) felt to be due to immaturity or
poor cooperation so were observed without refrac-
tion and were subsequently discharged without
treatment as their VA improved to 0.1 logMAR or
better. They were included as true or false positive
numbers at the time an accurate diagnosis was
made
An unknown number with mildly reduced vision
(see Supplement) were advised to go to the optician
and lost to follow up. From this cohort, which is of
a similar size to the ODS, 138 children with
confirmed reduced VA were referred by the ODS,
but only 83 were referred by the SNDS, suggesting
that many children were missed. If any with mildly
reduced VA went to an optometrist, none were
referred back for occlusion.
PPV was only 53%. Lack of data from optome-
trist referrals and the high false positive rate
reduces our confidence in the accuracy of the
data, and thus sensitivity and specificity, but they
are likely to be lower in this service.
Seventy-four children with confirmed reduced
vision were followed to discharge or the audit
date, and 8 still being occluded at the onset of the
2020 COVID lockdown, so final outcomes are not
available. Only two children still have VA in the
amblyopic eye of 0.4 or worse and both are still
being treated.
Fifteen (18.5%) of children with reduced VA
would have passed the AAPOS amblyopia risk fac-
tor referral threshold.
Costs
The actual screening costs cannot be identified
because the screening is part of a general health
screening, but they are carried out by personnel
on similar pay bands, so are likely to be similar.
The main difference in costs was in the higher costs
of the false positive referrals. Although costs per
genuine case and final VA outcomes for referred
children were similar, the cost of the 73 false refer-
rals was nearly seven times that of the other service,
while it is likely other children have been missed
and so remain untreated.
Assuming similar screening costs, the cost per
child with confirmed reduced vision seen at the
hospital was £285.70 (46% more); and per
amblyope detected was £948.09 (39% more), and
this does not consider the cost and outcomes of the
unknown number of children referred to local
optometrists.
Discussion
A single visual acuity screen at 4–5 years, led and
administered by orthoptists is unusual worldwide,
when many countries screen repeatedly, or earlier,
or use automated screening for refractive risk fac-
tors, rather than reduced VA itself.
10,11
The
Table 2. Actual costs at 2019–2020 published rates. Costs per
child of screening and postreferral treatment of SNDS and ODS
are likely to be similar.
COSTS (2019–2020 published rates) £ £
Exhange rate £1 = €1.12 = $1.25
Per screen (£) 4.21
Hospital appointment (£) 70
Per screen (£) 46.59
Box of patches (£) 8.75
Cost to amblyopia diagnosis (£) 189.8
Mean reduced VA treatment cost, diagnosis to
discharge (£)
331.68
Mean amblyopia treatment cost, diagnosis to
discharge (£)
458.65
ODS SNDS
Cost to service per child with low vision (£) 195.22 285.70
Cost to service per child with amblyopia (£) 683.28 948.09
Total cost of false referrals (£) 910.00 6300.00
STRABISMUS 5

UKNSC recommends that screeners should be
orthoptists, or those trained by orthoptists, and
this paper reports two ends of this spectrum –
orthoptist screeners vs. minimal orthoptist input
in two otherwise similar populations. The SNDS
training is not supervised by orthoptists as is
recommended. Data were much more accessible
for the ODS and full audit and feedback more
embedded, the importance of which is one of the
messages of this paper. Tight audit has led to many
small refinements of the ODS over the years, for
example, recalling children in school, and going
into the schools with a high proportion of disad-
vantaged or non-English speaking children later in
the screening cycle to allow them a better chance of
complying with testing because they had been in
school longer.
A single screening by an orthoptist followed
by orthoptist/hospital optometrist follow-up
appears to be highly efficient and very low cost,
both per screen, and per patient journey and per
amblyope detected, compared to other
alternatives.
12–14
It is often assumed that highly
skilled VA screening is more expensive,
5
but
these data suggest that it is not necessarily the
case. The current UKNSC recommendation to
not specifically test for strabismus, as is carried
out in many countries
11,15
seems supported. All
children with significant newly diagnosed stra-
bismus also had reduced VA.
The SNDS resulted in many more false referrals,
and therefore inefficient use of expensive hospital
services, and probably more false negatives. The
prevalence of refractive error is likely to be similar
in both areas, so more children with reduced vision
are probably missed. The ODS had a slightly lower
referral rate (3.41% vs. 3.82%), and the local
authority figures do not include the approximately
0.65% of the cohort (using the ODS criteria) who
might have been expected to be sent to local opti-
cians, not hospital services, so the true local author-
ity referral rate could have been nearer 4.5%.
The school nurses seem less expert in detecting
genuine reduced vision. The UKNSC recommends
that if orthoptists do not do the screening, they
should train and monitor the screeners very care-
fully, but that is not the case for the SNDS described
here. Orthoptists are particularly skilled in testing
children’s vision, but they too had to learn. If
orthoptists are not available, careful training to
a gold standard, regular supervised practice, feed-
back, and audit seem to be the issue that defines the
ability to test vision accurately, whoever does it.
A common argument for early or repeated
screenings is that early treatment leads to better
outcomes, and some cases of amblyopia can be
prevented. Children start compulsory schooling at
4–5 in England, so they are still well within the
critical period; but still much older than propo-
nents of earlier screening would like.
16
Nevertheless, of the 126 children with confirmed,
previously-undiagnosed reduced vision followed to
discharge, all improved with treatment, which was
usually glasses alone. 95% reached at least 0.2
logMAR acuity (the lower 95% CI of normal acuity
at age 6
6
) in each eye, and a few more have slowly
improving unilateral amblyopia. All children with-
out other pathology now have vision in at least one
eye well within normal ranges and there is only one
child with VA worse than 0.4 in an amblyopic eye
who could suffer any significant functional disad-
vantage (e.g. unable to drive a car) if they were to
lose their better eye. Children with strabismus
either present earlier,
17
would have been picked
up by VA testing, or have small or intermittent
strabismus with minimal amblyopia.
Photoscreening is often advocated because it can
be carried out earlier, by less skilled testers. But it
can result in high referral rates,
18
more appoint-
ments from screen to discharge and other potential
disadvantages (for review see Horwood et al.
4
)
Importantly, in this study, 14–18% of children
with reduced VA would have been missed by earlier
photoscreening.
Some might argue that those with 0.2 final
logMAR VA might have been ended up a line better
if treated earlier,
19
or that good acuity is necessary
for general development in the toddler years, but
none of these children’s parents had noticed any
visual problem, and concerned parents had already
sought treatment, so any disadvantage is likely to be
small. Parents sometimes report improvements in
behavior or performance after new glasses. This
evidence is often anecdotal and very prone to pla-
cebo effects, but does need further research. There
is little evidence that slightly blurred vision before
school entry holds children back, or that minimal
reduction in final best corrected acuity carries any
6A. HORWOOD ET AL.

lifetime disadvantage. In a public health context
such marginal gains may not be cost effective.
The main barrier to universal care was loss to
follow-up, and with the SNDS, legal and practical
barriers to data sharing. Data-protection laws pre-
vent health services contacting parents of screened
children directly, so if a parent does not return their
details, children cannot be sent appointments.
The main limitation of the study was data avail-
ability for the SNDS, but any COVID-related diffi-
culties in accessing some screening data do not
account for the differences in outcomes or costs.
Using estimates of false negatives is not ideal, but in
the case of the ODS they do seem realistic because it
is so rare to see any child that passed screening who
presents later with amblyopia. The orthoptists gen-
erally check back to screening records and remem-
ber doing so.
A further limitation is that the analysis did not
include the 2.3% of children who were wearing
glasses on the day of screening. When setting up
the service, it was decided not to test the vision of
these children as it was being professionally mon-
itored elsewhere, and any amblyopia would have
been detected and treated. This is an audit, not
a prevalence study, and the focus of the service is
the detection of previously undetected children at
minimum cost and maximum efficacy. We also did
not have the means to access these children’s diag-
nosis or prescription, and had no means to tell
whether they would have passed the screening
before they received the glasses (e.g. mild astigma-
tism or hypermetropia might be corrected by some
optometrists). We were therefore unable to analyze
their data further, and it does mean that the data
should not be used as a precise measure of preva-
lence in the population.
The point at which amblyopia is diagnosed can
vary. It could be on the day of refraction if VA does
not improve with lenses, on the first visit on return
with glasses or after full adaptation to glasses as
recommended by many research studies.
20,21
In
this study, we chose the most common clinical
definition (after 6–8 weeks of full-time glasses)
and many children had achieved equal VA at this
point.
It is possible that within the data there are differ-
ences between the communities, such as in socio-
economic or ethnic disadvantage, that could
explain some of the differences between the out-
comes. Although in the SNDS area there is one
town with a very high proportion of South Asian
and socio-economically disadvantaged families, it
sits in an otherwise largely White British, affluent
area, whereas in the ODS area the socio-economic
and ethnic groupings are more widely spread. On
average, the areas are broadly similar. While
screening and follow-up may need to address spe-
cific at-risk groups,
22
it was beyond the remit of this
audit.
These results strongly support the UKNSC
model as being highly cost effective. A single, accu-
rate screen by an experienced tester, at a site where
high coverage and a retest is possible, and a joined-
up service from screen to discharge seem key to
success. Training, feedback, and audit allow refine-
ments to improve services.
If more resources are to be allocated, it might be
better to use them to share best practice, improve
training, audit, quality assurance, follow-up atten-
dance rates, feedback, and local and national data
sharing, than add earlier, or more, screening events.
References
1. United Kingdom National Screening Committee.
Recommendation on vision defects screening in
children. London: Public Health England. https://legacy
screening.phe.org.uk/vision-child. 2013 Accessed April
09, 2021.
2. Simonsz HJ, Carlton J, Griffith H, et al. EUSCREEN
study, stage 1: data collection on vision and hearing
screening programs in 40 European countries and
Turkey, Israel, Russia, Malawi, Rwanda, Suth-Africa
and India. Invest Ophthalmol Vis Sci. 2019;60(9):3.
3. Solebo AL, Cumberland PM, Rahi JS. Whole-
population vision screening in children aged 4-5 years
to detect amblyopia. Lancet. 2015;385(9984):2308–2319.
doi:10.1016/S0140-6736(14)60522-5.
4. Horwood AM, Griffiths HJ, Carlton J, et al. Scope and
costs of autorefraction and photoscreening for child-
hood amblyopia—a systematic narrative review in rela-
tion to the EUSCREEN project data. Eye.
2020;35:739–752. doi:10.1038/s41433-020-01261-8.
5. Van der Ploeg CPB, Grevinga M, Eekhout I, et al. Costs
and effects of conventional vision screening and photo-
screening in the Dutch preventive child health care
system. Eur J Public Health. 2021 Feb 1;31(1):7–12.
doi:10.1093/eurpub/ckaa098.
6. Sonksen PM, Wade AM, Proffitt R, Heavens S, Salt AT.
The sonksen logMAR test of visual acuity: II. Age norms
STRABISMUS 7

from 2 years 9 months to 8 years. J AAPOS. 2008;12
(1):18–22. doi:10.1016/j.jaapos.2007.04.019.
7. NHS England national cost collection for the NHS.
https://www.england.nhs.uk/national-cost-collection/.
Published 2019. Accessed January 21 2021.
8. Donahue SP, Arthur B, Neely DE, et al. Guidelines for
automated preschool vision screening: a 10-year,
evidence-based update. J AAPOS. 2013;17(1):4–8.
doi:10.1016/j.jaapos.2012.09.012.
9. Stewart C, Moseley M, Fielder A, Stephens D, et al.
Refractive adaptation in amblyopia: quantification of
effect and implications for practice. Brit
J Ophthalmol. 2004;88(12):1541–1542. doi:10.1136/
bjo.2004.044214.
10. Carlton J, Griffiths H, Mazzone P, et al. Vision screen-
ing country reports. https://www.euscreen.org/vision-
screening-country-reports/. Published 2018. Accessed
August 07 2019.
11. Sloot F, Hoeve HL, de Kroon ML, et al. Inventory of
current EU paediatric vision and hearing screening
programmes. J Med Screen. 2015;22(2):55–64.
doi:10.1177/0969141315572403.
12. Konig -H-H, Barry J-C. Cost effectiveness of treatment
for amblyopia: an analysis based on a probabilistic
Markov model. Brit J Ophthalmol. 2004;88(5):606–612.
doi:10.1136/bjo.2003.028712.
13. Longmuir SQ, Pfeifer W, Leon A, Olson RJ, Short L,
Scott WE. Nine-year results of a volunteer lay net-
work photoscreening program of 147 809 children
using a photoscreener in Iowa. Ophthalmology.
2010;117(10):1869–1875. doi:10.1016/j.
ophtha.2010.03.036.
14. Donahue SP, Baker JD, Scott WE, et al. Lions Clubs
International Foundation core four photoscreening:
results from 17 programs and 400,000 preschool
children. J AAPOS. 2006;10(1):44–48. doi:10.1016/j.
jaapos.2005.08.007.
15. Pentland L, Patel S. Scottish pre-school vision screen-
ing – first 3 years of national data. Br Ir Orthoptic J.
2020;16(1):13–18. doi:10.22599/bioj.138.
16. Kirk VG, Clausen MM, Armitage MD, Arnold RW.
Preverbal photoscreening for amblyogenic factors and
outcomes in amblyopia treatment: early objective
screening and visual acuities. Arch Ophthalmol.
2008;126(4):489–492. doi:10.1001/archopht.126.4.489.
17. Telleman MAJ, Sloot F, Benjamins J, Simonsz HJ. High
rate of failed visual-acuity measurements with the
Amsterdam picture chart in screening at the age of 36
months. Acta Ophthalmol. 2019;97(1):24–28.
doi:10.1111/aos.13898.
18. Cordonnier M, Kallay O. Non-cycloplegic screening for
refractive errors in children with the hand-held autore-
fractor retinomax: final results and comparison with
non-cycloplegic photoscreening. Strabismus. 2001;9
(2):59–70. doi:10.1076/stra.9.2.59.701.
19. Stiff H, Dimenstein N, Larson SA. Vision screening out-
comes in children less than 3 years of age compared with
children 3 years and older. J Am Assoc Pediatr Ophthalmol
Strabismus {JAAPOS}. 2020;24(5):293.e291–293.e294.
20. Moseley M, Irwin M, Jones H, Fielder A, Auld R. Effect
of spectacle wear and minimal occlusion therapy on the
vision of amblyopic children. Invest Ophthalmol Vis Sci.
1996;37:s491.
21. Moseley MJ, Neufeld M, McCarry B, et al. Remediation
of refractive amblyopia by optical correction alone.
Ophthalmic Physiol Opt. 2002;22(4):296–299.
doi:10.1046/j.1475-1313.2002.00034.x.
22. Bruce A, Santorelli G, Wright J, et al. Prevalence of, and
risk factors for, presenting visual impairment: findings
from a vision screening programme based on UK NSC
guidance in a multi-ethnic population. Eye (Lond). 2018
Oct;32(10):1599–1607. Epub 2018 Jun 13. PMID:
29899459; PMCID: PMC6169729. doi:10.1038/s41433-
018-0146-8.
8A. HORWOOD ET AL.