A Paper of Advantages of the Sonksen Logmar test System over alternative Systems

PURPOSE To develop a standardized logMAR test of visual acuity for young children and establish
testability and reliability.
METHODS Two thousand nine hundred ninety-one children, aged 2 to 8 years 6 months, from a
population sample of 4671 were recruited from schools and preschool facilities and tested.
Ability to name or match letters, accept occlusion, and achieve binocular single and
binocular and monocular measures of linear visual acuity were recorded. Inter- and
intratester reliability was assessed for a random selection of 215 children.
RESULTS Two thousand nine hundred seventy-four (99.4%) matched or named letters, 2966 (99%)
achieved binocular single, 2940 (98%) achieved binocular linear, and 2807 (94%) achieved
monocular linear (for both eyes) measures. Binocular linear measures were achieved in
50% aged less than 3 years, 80% aged 3 to 3½, 92% aged 3½ to 4, and 99% aged 4 years
and over. Binocular linear measures were achieved in 86% of 2- to 3-year-olds and 96%
of 3- to 3½-year-olds who matched or named letters; monocular linear measures were
achieved in 86 and 95%, respectively, of these age groups who then accepted occlusion.
The median time for completion of the test ( binocular and monocular) was 3 minutes 55
seconds. The 95% limits of agreement for the same observer and for different observers
were 0.13 and -0.19, +0.15 for binocular linear and -0.16, +0.15, and -0.17, +0.13
for monocular linear measures, respectively.
CONCLUSIONS The Sonksen logMAR Test provides reliable binocular and monocular measures of linear
visual acuity in a high proportion of children from the age of 2.5 years. ( J

T
esting the visual acuity of young children in accord
with adult standards has posed problems to test
designers over several decades.
1-4
Many designers
have introduced stimulus modifications in attempts to sim-
plify the developmental content, for example, using sym-
bols instead of letters.
5,6,7
The Sonksen Silver Acuity System successfully ad-
dressed many of the developmental problems of letter-
based tests for young children, while adhering strictly to
the specifications of the Snellen standard, and was the first
to introduce single-line test displays in a flip-over book-
let.
8
In 1995 our group demonstrated high testability even
of the youngest children for this test and maturation in
visual acuity between 2 years 6 months and 9 years.
9,10
The Consilium of Ophthalmologists decided that the
logMAR scaling devised by Bailey and Lovie in 1976
had advantages over Snellen scaling and is now the
accepted gold standard.
11-14
When we decided to design
the Sonksen logMAR test, existing logMAR tests for
children were published without norms: in some the
spacing between optotypes was less than that specified
for adult tests, such as the ETDRS.
15-18
Such factors
compromise the interpretation of measurement in terms
of the adult standard and the accuracy of monitoring
changes in visual acuity due to treatment or progress of
disease from childhood into early adult life. A standard
logMAR test system together with testability, reliability,
and norms was therefore needed for preschool and
young school-age children.
In this article we present the first two stages in the
development of the Sonksen logMAR Test: (1) test
design and standardized protocol and (2) testability and
reliability. We present the third stage—the develop-
ment of age norms—in the companion article.
19
The
new test has logMAR scaling, uses international chart
design standard letter optotypes with constant contour
interaction, and uses a standard test protocol that in-

cludes features aimed at enhancing testability of the
youngest age groups.
Materials and Methods
Ethical approval was obtained from the Ethics Committees of
Cambridge District ( UK) and the Institute of Child Health
London. Data collection was organized and collated by the re-
search orthoptist and administrator. Five other community or-
thoptists assisted the research orthoptist with vision testing after
instruction in the standard test protocol (see e-Supplement 1,
available at jaapos.org); this ensured uniformity of test adminis-
tration throughout the study.
Design of the Sonksen logMAR Test
Developmental Features. In designing the Sonksen Sil-
ver Acuity System and the Sonksen logMAR Test, we adhered to
fundamental developmental principals of test design: when
adapting test design to suit an age or disability group, one
modifies/simplifies the instructions and response task to suit the
group while maintaining the fabric/standard of the test content.
The developmental features of the Sonksen logMAR Test and of
the test protocol are similar to those of the Sonksen Silver Acuity
System,
8
with a training booklet, keycard, one single-letter, and
two linear flip-over test booklets, pointer, and occlusion specta-
cles. Each page of the linear test booklets presents a single line of
four letters with crowding bars. The letters O, X, H, T, U, and
V were selected as these are the easiest for young children to
match.
2
A test distance of 3 m was chosen as this enhances
rapport and helps maintain the attention of the youngest children
without affecting contour interaction or decreasing sensitivity or
detection rate of myopia.
20,21
A flip-over booklet of crowded
single letters was used to facilitate choosing a suitable starting
level in the linear booklet as young children are likely to tire and
give erratic responses if asked to identify all the test letters on an
eye chart.
Test Specifications. The specifications recommended by
the Consilium of Ophthalmologists and others were closely ad-
hered to in the test design.
11,12,18,22
Optotypes and the spacing
between them conform to ETDRS specifications, that is, letters
are 5 minutes of arc square, without serifs, and with stroke width
equivalent to 1 minute of arc at every level. Spacing between
optotypes is equal to one letter width. Figure 1 illustrates a linear
test display. (For full design details, see e-Supplement 1 and the
Figure 1 legend.)
Test Protocol. Younger children are seated at a low table
and trained in letter matching using the training booklet and
keycard if unable to name letters. Children are then asked to
identify each of the letters in the booklet of single-letter displays
presented at 3 m starting with the 0.7 letter until a letter is not
seen correctly. The smallest letter seen is recorded as the child’s
binocular visual acuity for singles. The test display two logMAR
levels above the singles failure in the first linear test booklet is
then presented. If a child is unable to see some of the letters at
the starting level, the tester should present the preceding larger
displays until a full line is identified; the child is then shown each
letter in turn until three consecutive letters are failed. Every
letter correctly identified after the last whole line achieved con-
tributes to the score.
23
Occlusion spectacles are then introduced
and monocular acuity is ascertained using the second booklet of
linear displays for the right eye and the first booklet for the left
eye. The time taken for each phase of testing was recorded. (For
fuller details, see e-Supplement 1.)
Study Design. The design of the study was cross-sectional
and population based.
Recruitment. Children aged 2 years to 8 years 8 months
were recruited through Cambridge ( UK) city, state, and private
sector schools, nurseries, and playgroups. No visual defect or
impairment criteria were applied. Parents/guardians who gave
consent provided information on demographics and prior vision
treatment via a questionnaire. Class listings were used to ascer-
tain the total population from which the responders were drawn
and to check demographic data, including gender and postal
code. Ethnicity was recorded at the time of testing. Socioeco-
nomic status was based on the Townsend Index, which provides
a material measure of deprivation and disadvantage based on data
from the 2001 UK population census. Higher Townsend index
scores are associated with higher deprivation and disadvantage.
24
Testing. Ability to match or name letters, to accept oc-
clusion, and to achieve the four measures of visual acuity—
binocular singles, binocular linear, and monocular linear
(right eye then left eye)—were recorded together with the
measures achieved. Spectacles, if worn (4% of children tested),
were removed during testing. Patients who failed to meet
predefined visual acuity criteria were referred for ophthalmo-
logic evaluation.
Reliability. A random selection of 215 children were re-
tested within 11 days, 103 by the same orthoptist, to ascertain the
level of inter- and intratester reliability. These children came
from across the test sites to ensure a distribution of ages in line

The number of children wearing glasses or already re-
ceiving ophthalmic treatment was 195/3014 (6.5%), which
is similar to the number of expected ocular or vision
defects requiring treatment or follow-up in the general
preschool population.
27-29
For those who consented, there was no difference in the
presence of a known eye/vision problem or English as first
language between those tested and those who were absent
or “unwilling” to be tested on the day.
Testability
Table 1 presents testability with 95% confidence intervals
(CI) for each age group and the factors that influence
ability to complete the “test”—achieving binocular and
monocular linear visual acuity measures for both right and
left eyes. Of the total 2991 children tested, 2807 (94%)
completed the test. Almost all (2236/2270 [99%]) children
aged 4 years and older completed the test; 50% of children
under 3 years old, 80% of 3- to 3½-year-olds, and 92% of
3½- to 4-year-olds did so.
Developmental factors (ability to letter match and to ac-
cept occlusion) influenced the ability of children less than 4
years old to complete the test. Letter matching was achieved
in 94 to 100% of children from 2½ years. Of those able to
letter match, 87% (95% CI, 80-92) of children under 3 years
old, 96% (95% CI, 93-99) of 3 to 3½-year-olds achieved a
binocular linear visual acuity measure. Only 7 of the 2630
(0.3%) children aged 3½ and older who were able to letter
match did not complete the binocular linear test.
Acceptance of occlusion was lowest in children under
3½ years. Of those who had achieved a binocular measure,
73% (95% CI, 65-81) of under 3 years old, 91% (95% CI,
87-95) of 3- to 3½-year-olds, and 95% (95% CI, 92-97) of
3½- to 4-year-olds accepted occlusion. Subsequently
74/86 (86-95% CI, 79-93) of under 3-year-olds and 171/
181 (94.5-95% CI, 91-98) of 3- to 3½-year-olds who had
accepted occlusion achieved monocular measures. From
the age of 3½ years, more than 99% of those accepting
occlusion completed the test.
The median administration time, the range, and
interquartile range by age group is given in Figure 3. In
children unable to name letters, time taken included train-
ing to letter match. The median time for completion of a
binocular single and linear measures was 2 minutes, 1
second and for completion of both binocular and monoc-
ular testing was 3 minutes, 55 seconds. Even in children
younger than 3 years, the median time to complete both
binocular and monocular tasks was 6 minutes with a min-
imum of 3 minutes, 7 seconds.
Reliability
The Bland-Altman plots of within (A) and between (B)
orthoptist reliability are shown in Figure 4 for binocular
linear measures and Figure 5 for monocular linear mea-
sures— only right eyes were retested. For all measures
there was good agreement on average between repeat tests
for both the same and different observers. The limits of
agreement show the range within which 95% of differ-
ences are expected to lie. The average difference and limits
of agreement are shown on the plots together with 95%
confidence intervals. The confidence intervals show the
precision of the mean and limit estimates.
For the same observer, the mean difference between
repeat test scores for binocular linear visual acuity was
0.0024 logMAR (95% CI, -0.010-0.015) and the 95%
limits of agreement were 0.13 logMAR (5.2 letters).
For different observers the mean difference was -0.016
(95% CI, -0.032-0) and the 95% limits of agreement
were -0.19, +0.15 logMAR (-7.6, +6 letters). For the
same observer the mean difference between repeat test
scores for monocular linear visual acuity was -0.0082
(95% CI, -0.024-0.007) and the 95% limits of agreement
were -0.16, +0.15 logMAR (-6.4, +6 letters). For dif-
ferent observers, the mean difference was -0.0188 (95%
CI, -0.034-0.004) and the 95% limits of agreement were
-0.17, +0.13 logMAR (-6.8-5.2 letters).
There were no significant trends in differences with
average visual acuity level for single and linear binocular
and linear monocular measures ( p = 0.352, 0.106, and
0.269, respectively).
Regression analysis was used to determine whether the
differences in repeat measurements of single and linear
binocular and linear monocular visual acuity varied ac-
cording to whether the tested differed between repeats,
how long there was between repeats, the age of the child,
and/or their gender. The results for each of the four
factors are shown separately in Table 2. The p-values show

to symbols and are presented in single lines with constant
contour interaction. Other tests developed for pediatric
populations (eg, Kay and Lea symbol charts) use nonstand-
ard picture stimuli
1,30
or reduced spacing (increasing con-
tour interaction) of letters (eg, the Keeler logMAR Test)
7
compared with the adult standard. These variations make
accurate comparisons difficult as the child matures and
adult tests are used.
This large population-based study of children aged 2 to
8 years 8 months using the Sonksen logMAR Test shows
high testability for both binocular and monocular testing
even in children aged 2½ to 3 years. Sonksen introduced
features into the design of the Sonksen Silver Acuity Sys-
tem to reduce the developmental complexity of the test
task and proposed a test protocol that improved testability
in the youngest children.
8
We incorporated similar devel-
opmental features and standard test protocol into the
Sonksen logMAR Test and again found high levels of
achievement for monocular measures—51% of 2½- to
3-year-olds, 80% of 3- to 3½-year-olds, and 92% of 3½-
to 4-year-olds. The important developmental features of
the protocol are: (1) sitting the child at a nursery table and
chair, (2) using the techniques described to train matching
and level finding, and (3) taking a full binocular linear
visual acuity measure before attempting occlusion.
8,9
Ad-
ministration time for the test is very acceptable for a
screening or diagnostic program and similar to that re-
ported by others.
9,31,32
Most studies have either not included
33
or have reported
poorer testability than found in this study in under 3½-
year-olds.
34-38
Groups who have looked at testability in
3-year-olds often fail to report separately on 3- to 3½-
year-olds and 3½- to 4-year-olds. In this study testability
was considerably better in the older group, suggesting that
testability is likely to be falsely high in a group of 3-year-
olds if the distribution is weighted in favor of the older age
bracket.
The rationale for starting with a monocular test even in
young children has been the clinical priority to identify
amblyopia. Our experience and the findings of our current
and previous studies suggests that completing binocular
testing first familiarizes a child with the test procedure
before adding the additional demands of occlusion and
thereby increases success with monocular testing.
9
In this
study testability is higher in 2- to 5-year-olds than in
recent studies by others using protocols that start, as rec-
ommended by the Task Force
12
with instruction in letter
matching ( binocular pretesting) at 1 or 3 m.
30,31,36
The
clinical value of full binocular testing is highlighted by the
findings of two large epidemiologic surveys of visual acuity
in childhood and two evaluation studies, which show that
little reliability in detection of uniocular problems is
lost in patients without strabismus, by achieving a bin-
ocular measure only.
27-29,39
On the basis of the above
evidence and as developmental experts with a major
interest in vision, we advocate that a full binocular
measure is undertaken before attempting occlusion or
monocular testing in under 5-year-olds.
The 95% limits of agreement of logMAR scores for the
Sonksen logMAR Test found in this study for monocular
visual acuity for the same (-0.16, 0.15) and different
observer (-0.17, 0.13) was in keeping with that previously
reported for logMAR-based tests in adult subjects and in
older children (5 to 9 years) (+0.14, -0.14 to +0.2, -0.2),
despite the inclusion of young children in this co-
hort.
12,40-42
A clinically relevant change in visual acuity
would from these results have to fall outside the 95%
limits of agreement, that is, for two measures made by the
same observer, a clinically relevant difference would be
that deemed to be outside the expected differences that
would be observed between repeats anyway: more than five
letters for binocular visual acuity and more than seven
letters for monocular visual acuity.
Differences in repeat measures were not significantly
influenced by any of the factors examined. The most
marginal results were for age; however, the point estimates
show that any differences are small. For example, the
estimated change with age for single binocular repeats is
negative, indicating better agreement for older children as
would be expected, yet the fall is only estimated to be
0.00098 per year and even over the entire age range in-
vestigated (2-9 years) this is an estimated change of less
than 0.007 (less than one letter). On the basis of these
results, it is reasonable to infer that clinically important
differences did not occur. Although the sample sizes are
not large for some subgroups, the coefficients and confi-

dence intervals suggest that clinically important differ-
ences were not missed due to a lack of power.
The strengths of this study are the adherence to inter-
national test design standards, an acceptable response rate
(70%) with consent only withheld actively in 77 (1.6%),
and the size and representativeness in terms of ethnicity
and socioeconomic status of the population-based cohort
to that of the UK.
Limitations include lack of demographic data concern-
ing nonresponders as demographic details were only col-
lected with the returned consent. The potential bias of
parents of children with visual difficulties being less or
more likely to respond is unlikely given the prevalence of
defects found in the population tested.
Conclusions
This article presents the first two stages in development of
the Sonksen logMAR Test, a new test for children that
accords with the adult standard and with a standard test
protocol. Levels of testability are higher and reliability
data are comparable to other tests. Stage 3, the derivation
of age norms, is presented in the companion article.