Autorefractors in Preschool Screenings

1
Autorefractors in Preschool Screenings
W.R. Bobier, L. Cowen, C. Machan, M. Lane, M.
Parks, D. Wintermeyer, B. Robinson
School of Optometry, University of Waterloo,
Waterloo, Ontario, Canada
Introduction
Results
Pre School Child enters vision screening study

Hand-held autorefractors such as the Retinomax
(Nikon Co) have been suggested for use in
preschool screenings. Best results appear to be
when cycloplegia is used.1,2 For large scale
preschool screenings however, non cycloplegic
measures are required. It is not clear if hand
held autorefractors can control accommodation
sufficiently in preschool children.3 We look at
the performance of two autorefractors, (Figs. 1
2) the Retinomax Plus (Nikon Co) and two DAV
prototypes (SureSight, Welch Allyn Co) employed
in a County wide vision screening programme for
preschool children. The DAV prototype uses a more
remote working distance (40cm) than the
Retinomax.
Preschool Children The mean equivalent sphere (M)
and cylinder components (J0, J45) for the
practitioners retinoscopy were 0.71D, 0.13D and
-0.01D respectively (Table 1). The difference
between the retinoscopy and the Retinomax and DAV
prototypes are shown in Table 2. It can be seen
that the equivalent sphere of the Retinomax is
significantly less hyperopic with respect to the
retinoscopy. However, cylinder values show no
significant departure.

• Subjective tests of visual and stereoacuity
  Pass/Fail
• Objective measure with Welch Allyn SureSight
  (DAV) auto-refractor (data only)
• Objective measure with Nikon Co. Retinomax Plus
  auto-refractor fail gt 2.00 D Sphere

Child fails one or more screening criteria
Table 5. Over Correction to match instrument with
retinoscopy. Conversion of the differences in M,
J0, J45 back to negative sphero-cylinder form.
Child passes screening
Adults The results for the Retinomax and DAV2
show the adult sample to be hyperopic in Table 6.
Retinomax measured 16 of the equivalent spheres
less than -1D, whereas Dav2 measured 65.
Percentages greater than 1D were 4 and 2
respectively. J0 components are significantly
different from the practitioners measures, but
only Retinomax has a significant J45 component.
An analysis of variance, split-plot design was
used to test the hypothesis that variation in the
preschool and adult populations was equivalent
when considering population as a factor.
Population does have a significant effect
however, only when considering the equivalent
sphere component (F168.07, p0.0001). This
would imply that differences in cylinder
components are not coming from an effect of the
population.

• Referred to Eye Care Practitioner
• Specific clinical measures taken
• Form completed and sent to Oxford County Public
  Health Unit

Discharged from the study
Table 1. Means of M, J0 and J45 components, for
eye-care practitioner, right eye of the Preschool
sample.

• Screening Data sent to Waterloo
• Data Analysis

• Clinical Results collected by Oxford County
  Board of Health
• Clinical data sent to the University of
  Waterloo and combined with screening data
• Data Analysis

Figure 3. Flow Chart of Oxford County Vision
Screening Program
Figure 1 THE NIKON RETINOMAX PLUS
Adult Study
A separate investigation of the two instruments
was conducted on an older population with mean
age 38.7 years, range (7 to 78). Subjects who
gave prior consent were selected from a clinical
population at the School of Optometry, University
of Waterloo. A total of 167 subjects were
tested. All retinoscopic measures were reviewed
from the clinic record and classified in terms of
close agreement with the subjective refraction.
Cases where the discrepancy exceeded 0.50D were
not included. Instrument performance was
determined by comparison with retinoscopic
measures taken from the practitioners reports.
Table 6. Mean difference in refractive error
components between the practitioner and Retinomax
and the practitioner and DAV2 in an adult sample.
Table 2. Differences in means between
practitioner and instrument measurements of M, J0
and J45 (right eye). Paired t-test significance
levels for the preschool children.
A recent study of 69 adult subjects showed that a
third SureSight prototype (DAV3) had much smaller
biases with respect to the retinoscopy (M -0.58
J0 0.04 J45 0.01). Only the equivalent sphere
difference was significant.
DAV prototypes are seen to be more hyperopic with
respect to the retinoscopy. Further, the J0
component in both DAV prototypes is significantly
different from those of the practitioner while
only DAV1 has a significant J45 component. Final
analysis of the larger sample of 115 showed no
significant changes from the smaller
sample. DAV1 and DAV2 measures were taken on 137
of the children. The DAV1 and DAV2 prototype
measures could not be pooled since analysis of
variance testing showed that the prototypes were
significantly different in measures of M
(F7.01, p0.001) and J45 (F35.17,
p0.0001). Counts of equivalent spheres that are
1 D greater or less than those of the
practitioner for the initial sample (n88) are
given in Table 4. As expected with the bias for
the Retinomax was less hyperopic and the DAV
prototypes were more hyperopic in both eyes.
Conclusions
Data Analysis
Retinomax measures of refractive errors in
preschool children are more myopic than clinical
retinoscopy. We conclude that the close working
distance of the Retinomax induces variable
instrument myopia in non-cyclopleged preschool
children despite a fogging target. This is not
found in the adult sample. However, the
Retinomax provides a robust measure of
astigmatism in preschool children. The DAV
prototypes have reduced over-accommodative
effects. This is likely due in part to their 40
cm working distance. However, they show similar
variances in measures. The cylinder discrepancy
in the early versions has been significantly
improved in the latest prototype. The
accommodative behaviour of the preschool children
appears to lead to considerable variance in
autorefractor and possibly clinical measures.
Figure 2 THE WELCH ALLYN DAV2 PROTOTYPE
Data analysis was conducted by decomposing
sphero-cylinder values using a Fourier series
representation that makes refractive errors more
amenable to traditional statistical analysis.5
Specifically, the sphero-cylinder components are
decomposed into independent terms representing
the equivalent sphere, (M) and two
cross-cylinders components one at axis 0? (J0)
and the other at axis 45?( J45). The negative
sphero-cylindrical format is specified as (S, C x
?), (where C is a negative number) and is
transformed as follows
Preschool Study
A screening program, targeting all kindergarten
registrants, is conducted annually by the Oxford
County, Health Unit (Ontario, Canada).4 Parental
consent is obtained for inclusion of each child
in the study. During the spring of 1999,
vision-screening tests were performed on 1180
children (mean age 46.9 months, range 39 to
62). Of these, 369 children were referred to eye
care practitioners who were most predominantly
optometrists. Screening was primarily based on
subjective measures of visual and stereo acuity.
The pass/fail criteria was 6/6 acuity (single
letters) and 100 sec stereoacuity (SteroFly,
Titmus Optical Co). When a child was referred,
the eye-care practitioner reported specific
ocular-visual findings to the county health unit.
Figure 3 outlines the screening and data
collection.The autorefractors were introduced as
part of the screening but did not affect the
pass/fail decision. The sole exception was for
hyperopes showing 2D or more with the Retinomax.
The refractive measures of the two instruments
were compared to the retinoscopic measures
reported by the eye-care practitioners.Only data
with sufficiently high confidence levels were
accepted. The initial analysis was conducted from
a sample of 92 children referred, followed by an
analysis on an increased sample of 155 children.
Of the 92 initial children, 19 received a
cycloplegic refraction.
Acknowledgements
The support of the Oxford County Public Health
Board and the assistance of Ms. Kathleen Hill are
greatly appreciated. This research was funded by
a grant from the Welch Allyn Co.
References 1 Harvey EM, Miller JM, Wagner LK,
Dobson V. 1997. British Journal of
Ophthalmology 81(11) 941-8. 2 Cordonnier M, and
Dramaix M. 1998. British Journal of
Ophthalmology, 83(2) 157-61. 3 Wasserman W, and
Dick B. 1997. Klin Monatsbl Augenheilkd, 211(6)
387-94. 4 Robinson B, Bobier W, Martin E, and
Bryant L. 1999. American Journal of Public
Health, 89(2) 193-198. 5 Thibos L, Wheeler W,
and Horner D. 1997. Optometry and Vision Science,
74(6) 367-375.
Table 4. Count of the measurements that are
greater than or less than 1 Diopter from the
practitioner measurements of equivalent sphere M
(e.g. Mpractitioner Mretinomax)
Conversion of the instrument bias defined by M,
J0 and J45 back into sphero-cylinder form is
shown in Table 5. This represents the over-
correction required to match the instrument with
the retinoscopy.
Instrument differences (bias) could be determined
by subtraction from the clinical retinoscopy.
Statistical significance was determined using a
paired t-test.