Sensitivity of Photoscreening to Detect
High-Magnitude Amblyogenic Factors
Sean P. Donahue, MD, PhD,a,b,c Tammy M. Johnson, MPH,a Wanda Ottar, CO, COMT,d and
William E. Scott, MDd
Purpose: To determine the sensitivity of a unique pupil-size based set of referral criteria of the MTI PhotoScreener
(Medical Technology and Innovations, Inc, Cedar Falls, Iowa) to detect high magnitude refractive error. Methods:
The photoscreening photographs of 949 preschool children previously analyzed were reevaluated with the new
referral criteria. The original photographs had been obtained from pediatricians’ offices and public health and
Women, Infants, and Children’s (WIC) clinics. The results of this analysis were compared with the gold standard
clinical examination and cycloplegic refraction. Sensitivities were calculated for amblyogenic factors based on the
magnitude of the refractive error. Results: For 26 patients with anisometropia, the sensitivity to detect anisometropia increased from 46% for ⫹1.25 or greater spherical interocular difference to 100% for ⫹2.50 spherical
intraocular difference. For 36 patients with hypermetropia in at least 1 meridian ranging from ⫹3.75 to ⫹7.50 D,
sensitivity increased from 53% to detect ⫹3.75 D or greater to 70% for ⫹5.00 D or greater. The sensitivity to detect
hypermetropia of ⫹5.75 D or greater was 100%. These criteria detected 82% of patients with astigmatism greater
than or equal to ⫹3.00 D, and 100% of patients with astigmatism greater than ⫹3.50 D. Conclusion: It is crucial
that screening programs avoid overreferrals caused by high false-positive screening rates. The sensitivity of our
new criteria increases with higher magnitude refractive error; patients with moderate and severe amblyogenic
factors are almost never missed. While the sensitivity to detect lower magnitude refractive error is poor, the
amblyogenic impact of such errors remains to be determined. (J AAPOS 2002;6:86-91)
hotoscreening is receiving increased acceptance as a
method of preschool vision screening in spite of
legitimate concerns regarding interpretation of
photographs and overreferral rates.1,2 The MTI PhotoScreener is an eccentric, off-axis photoscreener that has
been tested extensively in various settings3-7 and the field.8
The MTI PhotoScreener detects amblyogenic factors
rather than amblyopia, but shows promise for preschool
vision screening.
Many studies investigating photoscreening with the
MTI PhotoScreener have reported sensitivity and specificity values.1-7 In all these studies, the reported sensitivity
and specificity depend on the criteria to judge a photo-
P
From the Departments of Ophthalmology,a Neurology,b and Pediatrics,c Vanderbilt University School of Medicine, Nashville, Tennessee, and the Department of Ophthalmology,d
University of Iowa Hospitals and Clinics, Iowa City, Iowa.
Presented at the 27th Annual Meeting of the American Association for Pediatric Ophthalmology and Strabismus, Orlando, Florida, March 22, 2001.
Supported in part by Research to Prevent Blindness, Inc, New York, New York. Dr
Donahue is the recipient of a Career Development Award from Research to Prevent
Blindness, Inc.
The study was conducted at Vanderbilt University Medical Center.
Submitted April 2, 2001.
Revision accepted October 19, 2001.
Reprint requests: Sean P. Donahue, MD, PhD, Department of Ophthalmology and Visual
Sciences, Vanderbilt University School of Medicine, 8000 Medical Center East, Nashville,
TN 37232-8808.
Copyright © 2002 by the American Association for Pediatric Ophthalmology and
Strabismus.
1091-8531/2002/$35.00 ⫹ 0 75/1/121168
doi:10.1067/mpa.2002.121168
86
April 2002
graph abnormal (referral criteria) and the criteria to call a
clinical examination finding abnormal (failure criteria) in
addition to the person interpreting the photograph. Because photoscreening detects amblyogenic factors rather
than amblyopia, and since many children with amblyogenic factors may never develop amblyopia, criteria both
for referral and failure may differ significantly between
studies but still be valid, depending on the screening situation. Nevertheless, a universal factor for all referral
criteria should be a progressively increasing sensitivity to
detect amblyogenic factors of higher magnitude, since
they are more likely to be amblyogenic.
We have been developing and refining a volunteer-led
vision screening program for preschool children. Our
statewide screening program uses lay volunteers from local
Lions Clubs, a photograph reading center, and a network
of optometrists and ophthalmologists in the child’s local
community. Because we were concerned that a high referral rate and the resulting high false-positive rate could
jeopardize the reputation of this volunteer-led program in
the minds of community physicians, we developed unique
criteria for referral. These criteria produced a referral rate
of approximately 5.4% and a predictive value positive of
approximately 70%, after screening 45,000 preschool children. Details regarding this program have been published
elsewhere.8
Although we have published data regarding the referral
rate and predictive value positive for these referral criteria,
Journal of AAPOS
Journal of AAPOS
Volume 6 Number 2 April 2002
Donahue et al
TABLE 1. Vanderbilt University photoscreening referral criteria
1. Any strabismus
2. Media opacity ⬎1 mm
3. Refractive error for:
4 mm pupillary diameter
6 mm pupillary diameter
a. Myopic crescent ⱖ 2.5 mm
Myopic crescent ⱖ 4.5 mm
b. Hyperopic crescent ⱖ 2.5 mm
Hyperopic crescent ⱖ 4.5 mm
c. Astigmatism ⬎ 1.5 mm
Astigmatism ⬎ 2.0 mm
d. Anisometropia: No crescent in fellow eye:
Crescent ⱖ 2.0 mm
Crescent ⱖ 3.5 mm
e. Anisometropia: Crescent in fellow eye:
ⱖ1 mm crescent in fellow eye
1) ⱕ2.5 mm crescent in fellow eye
1 mm difference between eyes
2 mm difference between eyes
2) ⱖ 3 mm crescent in fellow eye
1 mm difference between eyes
we have not reported sensitivity data. Obtaining such data
with adequate confidence intervals would require comprehensive eye examinations with cycloplegic refraction on
large numbers of children, and our program does not have
the resources to do that. Instead, we obtained the photoscreening photographs from 949 previously reported children, who were photoscreened and who had a comprehensive eye exam, and whose results were used in determining
initial sensitivities for the MTI PhotoScreener.3 We
sought to determine the sensitivity of our referral criteria
to detect various types of amblyogenic factors. We also
determined the sensitivity of these criteria to detect refractive error of increasing magnitude.
METHODS
Subjects
Data regarding the subjects used in this study have been
reported previously.3 They come from a data set of 1003
healthy children between the ages of 6 months and 59
months who were examined during routine vision screening at Public Health, and Women, Infants, and Children’s
(WIC) clinics in Iowa and Illinois, and from children
evaluated in private offices of pediatricians in Iowa and
Texas. Although the relative distribution of pathology for
these children had been reported, in this study, the photographs were evaluated by trained readers in a masked
fashion without having access to the results of the comprehensive eye examination and cycloplegic refraction.
(The staff at the Vanderbilt Ophthalmic Imaging Center
(VOIC) was not familiar with photoscreening literature
and had not read the previous report by Ottar et al.3)
All photographs were evaluated at the VOIC. Photographs were first determined to be readable or unreadable.
Unreadable photographs had a pupil size smaller than 4
mm or lack of fixation on the camera. Once the determination of a readable photograph was made, the referral
criteria shown in Table 1 were applied. All photographs
were first interpreted by VOIC staff and then by 2 of the
authors (S.P.D. and T.M.J.). A consensus of all 3 was
necessary before a final grading was given. (Currently all
87
8 mm pupillary diameter
Myopic crescent ⱖ 6.5 mm
Hyperopic crescent ⱖ 6.5 mm
Astigmatism ⬎ 2.5 mm
Crescent ⱖ 4 mm
1) ⱕ3.5 mm crescent in fellow eye
2 mm difference between eyes
2) ⱖ 4 mm crescent in fellow eye
1 mm difference between eyes
photographs that are referred or considered unreadable by
VOIC staff are also reviewed by both S.P.D. and T.M.J.,
and a consensus is required prior to making a formal
reading.) When several amblyogenic factors were present,
the amblyogenic factors were ordered as to the most significant diagnosis (media opacity ⬎ strabismus ⬎ anisometropia ⬎ hyperopia ⬎ myopia ⬎ astigmatism). Anisometropia was determined by comparing across meridians.
(When astigmatism was present in only 1 eye, it was
considered to be anisometropia). Because only 3 patients
had strabismus on the formal exam, these were later excluded from analysis except for their refractive errors. No
patients had media opacity. No patients had myopia without also having superimposed anisometropia or astigmatism. Therefore, for this study patients were categorized
into 3 groups based on the primary refractive problem: (1)
anisometropia, (2) hypermetropia, and (3) astigmatism, in
that order of diagnosis. In addition to the sensitivity for
the primary diagnosis, we also calculated sensitivities for
patients with anisometropia, hypermetropia, or astigmatism without regard to the primary diagnosis (ie, a patient
with anisometropia and hyperopia is given a primary diagnosis of anisometropia, but is also included in the hyperopia group). It should be noted that the flash on the
MTI photoscreener is oriented so as to detect meridional
astigmatism; a significant limitation of this instrument is
the relative inability to detect oblique astigmatism. We
also determined the sensitivity to detect various magnitudes of each amblyogenic factor. For purposes of classification, we used the criteria for examination failure listed
in Table 2. These criteria were derived from a survey of
pediatric ophthalmologists9 and were the exam failure criteria used for previous reports regarding this program.8,10
TABLE 2. Criteria for failure of clinical eye examination
1. Anisometropia ⬎ 1.00 D
2. Astigmatism ⬎ 1.50 D
3. Hypermetropia ⬎ 3.50 D
4. Myopia ⬎ ⫺3.00 D
88
Journal of AAPOS
Volume 6 Number 2 April 2002
Donahue et al
TABLE 3. Sensitivity to detect anisometropia
Amount*
N
Passed
Referred
⫹1.25
⫹1.50
⫹2.00
⫹2.50
⫹3.00
⫹3.50
Total
Sph
Sph
Sph
Sph
Sph
Sph
5
12
6
1
1
1
26
3
10
1
0
0
0
14
Sensitivity†
2
2
5
1
1
1
12
46%
48%
89%
100%
100%
100%
*Inter eye difference in refractive error in greatest meridian.
†Sensitivity to detect amblyogenic factor of this or greater magnitude.
RESULTS
Photo Interpretation
The original Iowa data set had 978 patients. We excluded
39 photographs because of a small pupillary diameter or
difficulty obtaining a photoscreening photograph or clinical information. Of the 939 patients with interpretable
photographs and examination results, 88 had significant
abnormalities. Twenty-six met the clinical criteria for primary anisometropia (⬎1.00 D), 40 met the clinical criteria
for primary astigmatism (⬎1.50 D) and 22 met the criteria
for primary hypermetropia (⬎3.5 D in 1 meridian). However, because many patients with anisometropia also had
hypermetropia or astigmatism in 1 meridian, and many
patients with astigmatism had significant hypermetropia in
1 meridian, several patients could be categorized into multiple diagnostic groups. When classified in this manner, 51
children had astigmatism, 36 had hypermetropia, and 26
had anisometropia. Overall, the sensitivity for amblyogenic factors of all magnitudes (Table 2) was 50% and the
specificity was 98.5%. The negative predictive value was
94.4% and the positive predictive value was 57.5%.
Sensitivity to Detect Anisometropia
Twenty-six patients had anisometropia greater than 1.00
D detected on the clinical examination and cycloplegic
refraction. The distribution of anisometropic refractive
errors is listed in Table 3. Of the 5 patients with anisometropia of 1.25 D, 3 were passed by the VOIC while 2
were referred. Of the 12 patients with ⫹1.50 D anisometropia, 10 were passed and 2 were referred. One patient
was referred for suspected myopia and had it confirmed on
the examination (along with the anisometropia), and 1
patient was referred with suspected hypermetropia.
For the 9 patients having anisometropia of 2 D or
greater, sensitivity improved substantially. Six patients had
anisometropia of 2 D. Of these patients, 5 were referred
and only 1 was passed by the VOIC. The patients with
⫹2.50 D, ⫹3.00 D, and ⫹3.50 D anisometropia were all
appropriately referred.
In summary, the sensitivity to detect 1.25 D or greater
anisometropia was 46%. However, the sensitivity to detect
2.00 D or greater anisometropia was 89%.
TABLE 4. Sensitivity to detect hypermetropia
Amount (D)
N
Pass
Refer
⫹3.75
⫹4.00
⫹4.25
⫹4.50
⫹5.00
⫹5.25
⫹5.50
⫹5.75 –
⫹6.00
⬎ ⫹6.00
Total
Sensitivity
2
6
1
7
3
3
8
3
1
3
1
6
1
1
4
0
1
3
0
1
2
2
4
3
53%
53%
54%
55%
70%
71%
71%
100%
3
36
0
17
3
19
100%
*Sensitivity to detect amblyogenic factor of this or greater magnitude.
Sensitivity to Detect Hypermetropia
Thirty-six patients met the clinical criteria for hypermetropia (⬎3.5 D). Of these, 14 patients had astigmatism
or anisometropia, with hypermetropia in 1 axis of 1 eye
greater than 3.5 D, while 22 had hypermetropia without
anisometropia or astigmatism. Of these 36 patients, 17
were passed by the VOIC, whereas 19 were referred (Table 4). The 17 patients who were passed had hypermetropia ranging from ⫹3.75 D to ⫹5.50 D. The 19
patients who were referred had hypermetropia ranging
from ⫹3.75 D to ⫹7.50 D.
The sensitivity to detect ⫹4.00 D or more hypermetropia in at least 1 meridian was 53%. However, the
sensitivity to detect ⫹5.00 D or greater of hypermetropia
was 70%, while the sensitivity to detect greater than ⫹5.50
D of hypermetropia was 100%. Six patients with refractive
error of at least ⫹5.00 in 1 meridian were missed (falsenegatives).
Patients having hypermetropia along with astigmatism
or anisometropia were more likely to be detected than
those with hypermetropia alone. Of the 8 patients with
hypermetropia and astigmatism (hyperopia ⬎ ⫹3.50 D in
at least 1 meridian), 5 were referred. Of 6 patients with
anisometropia and hypermetropia, 4 were referred. However, this may reflect a sampling bias as patients with
hypermetropia and either astigmatism or anisometropia
had higher levels of hypermetropia in 1 axis than did those
with spherical symmetric hypermetropia. Sample size limitations precluded us from making this comparison more
directly.
Sensitivity to Detect Astigmatism
Forty patients met the clinical criteria for primary astigmatism (⬎1.50 D in 1 meridian without concomitant anisometropia or hyperopia). An additional 11 patients had
astigmatism, but with a primary diagnosis of anisometropia (n ⫽ 11). Thus, 51 children were considered to have
astigmatism for the purpose of this study. Of these 51
children, 29 were referred by the VOIC and 22 were
passed. The distribution of the astigmatism is listed in
Table 5.
Journal of AAPOS
Volume 6 Number 2 April 2002
TABLE 5. Sensitivity to detect astigmatism
Amount of
astigmatism
N
Pass
⫹1.75
⫹2.00
⫹2.25
⫹2.50
⫹2.75
⫹3.00
⫹3.50
⫹4.00
⫹4.50
⫹6.00
Total
11
13
1
14
1
3
4
1
2
1
51
7
7
0
5
1
1
1
0
0
0
22
Donahue et al
Refer
Sensitivity
4
6
1
9
0
2
3
1
2
1
29
57%
63%
70%
69%
75%
82%
88%
100%
100%
100%
*Sensitivity to detect factor of this or greater magnitude.
As the magnitude of astigmatism increased, the criteria
became more sensitive to detect it. The sensitivity to
detect astigmatism greater than 1.50 D was 57%, but
increased to 70% for astigmatism greater than 2.00 D.
The sensitivity to detect 3.00 D or greater of astigmatism
was 82%. These criteria detected 7 of the 8 children with
astigmatism of ⫹3.50 D and all 4 children with astigmatism greater than or equal to 4.00 D (100%).
There was no difference in sensitivity for the 40 patients with primary astigmatism (23 referred) compared
with the 11 with astigmatism associated with anisometropia (6 referred).
DISCUSSION
Screening for amblyogenic factors is inherently different
than most other types of health screening, because photoscreening identifies risk factors for a disease rather than
the disease itself. It is more similar to using cholesterol
levels to detect heart disease than it is to using mammography to detect breast cancer. Photoscreening must have a
high sensitivity to detect the highest magnitude amblyogenic factors, but also needs to have a low false-positive
rate. The perils and excessive costs associated with high
false-positive referral rates in photoscreening have been
well-documented by Simons et al.1
When we designed our program, we were keenly aware
of the effect that a high false-positive rate would have on
the reputation of a volunteer-led program with respect to
primary care physicians, parents, and the volunteers themselves. Our pilot studies had demonstrated a referral rate
that approached 15% when the manufacturer’s criteria
were used to interpret MTI photographs. In addition, we
recognized that in-the-field interpretation of photoscreening photographs by lay volunteers would present a massive
quality control problem. All of these concerns have been
documented since that time.1,2 Therefore, we designed a
unique grading system that used a pupil-based referral
criteria and a centralized reading center to deal directly
with these issues.8
Our photoscreening program has screened more than
45,000 preschool children (ages 6-48 months) in the state
89
of Tennessee using the MTI PhotoScreener, lay volunteers, and optometrists and ophthalmologists in the children’s local areas. Our current rate of follow-up has averaged more than 70%, and for the last 6 months of calendar
year 2000 exceeded 90% (For comparison, the immunization rate in Tennessee is 74%). Our current referral rate
averages 5.4% and our predictive value positive is nearly
70% (a 30% false-positive rate). However, in spite of our
high predictive value and low referral rate, we did not
know how well our criteria detected high magnitude refractive errors compared with those of moderate magnitude. Rather than perform such a study in the field, which
would require significant resources, we applied our criteria
to a previously studied group of children who had photoscreening photographs done using the MTI PhotoScreener and cycloplegic refractions by a trained orthoptist.
Our results demonstrate that while we have only fair
sensitivity to detect small magnitude refractive error, the
sensitivity increases with the magnitude of the refractive
error. In our study, no children with anisometropia greater
than 2.00 D, no patients with astigmatism greater than
3.50 D, and no patients with hypermetropia greater than
5.50 D were missed. We recognize that the small sample
size for these high refractive error groups limits conclusions about the exact sensitivity for these groups, but are
confident that these criteria detect the vast majority of
potentially amblyopic children without producing a large
number of false-positive referrals. Clearly, further studies
need to investigate larger numbers of patients with higher
magnitude refractive error.
Most patients with anisometropia of 1.50 D or less were
not detected using these criteria. However, it should be
noted that anisometropia of this magnitude is likely not
associated with more than a mildly increased risk for the
development of amblyopia. Kutschke et al11 found that
only 6 of 124 patients with anisometropic amblyopia had
anisometropia less than 1.5 D, and all of these had manifest strabismus. Because the patients in this study with low
levels of anisometropia were all orthotropic, it is not likely
that they will develop amblyopia. Ottar et al3 noticed an
increased sensitivity of the manufacturer’s criteria to detect larger magnitude anisometropia in this patient group;
those criteria detected 84% of anisometropia greater than
1.5 D.
Moderate degrees of hypermetropia were not detected
reliably under the noncycloplegic conditions of photoscreening. While this is concerning given the increased
risk of strabismus developing in patients with uncorrected
hyperopia,13 the natural history of moderate levels of hypermetropic refractive error in children who do not develop accommodative esotropia is unknown. We believe
that a child with mild to moderate hypermetropia who
accommodates without developing strabismus is likely at
minimal risk for developing amblyopia. In contrast, the
child with very high hypermetropia (⫹6.00 D or ⬎) is at
90
Donahue et al
greater risk of developing either accommodative esotropia
or refractive amblyopia, and must be detected with screening. We detected all 6 children with hypermetropia in any
meridian greater than ⫹5.50 D.
Tong et al12 have recently proposed guidelines for the
detection of hypermetropia with the MTI PhotoScreener
and we are currently prospectively comparing their guidelines to those reported here. Most of the false-negative
hyperopic children in this study were likely missed because
they remained orthotropic while accommodating partially
or fully on the camera fixation light. This problem has
been discussed previously by Ottar et al.3 A significant
problem with photoscreening for hypermetropia is that, in
the absence of accommodation, crescents become so large
with ⫹3.00 D of uncorrected hypermetropia that additional increments of hypermetropia produce little additional incremental increase in crescent size.14 It is unknown whether children who fail to accommodate this
degree at the time of photoscreening are at greater risk for
developing amblyopia than those who do accommodate.
Further research into mechanisms to detect latent hypermetropia with photoscreening and into the natural history
of moderate degrees of hypermetropia on the development
of amblyopia are warranted.
It is somewhat surprising that we were able to detect
moderate levels of astigmatism so well. This is even more
surprising in that similar levels of hypermetropia could not
be detected. We believe this is likely due to our criteria for
the detection of astigmatism, which require smaller crescents than do those for hypermetropia. However, a tradeoff of this increased sensitivity is a decrease in predictive
value positive; our predictive value for the suspected diagnosis of astigmatism is in the range of 42%, whereas that
for hypermetropia approaches 80%.7 An additional limitation of photoscreening with the MTI PhotoScreener is
that it is theoretically unable to detect oblique astigmatism
bilaterally (unilateral oblique astigmatism will be revealed
as anisometropia). No patients in this study had significant
oblique astigmatism, because of its rarity. Nevertheless, an
optimal screening instrument must be able to detect it.
The biggest limitation of this study is that the study
group refractions were performed by the same person who
did the photoscreening. This represents a potential bias
that could not be overcome without reproducing the entire study. We interpreted all photographs, however, prior
to receiving examination results on these children and
therefore were masked to their examination findings at the
time of photograph interpretation.
Many sets of criteria have been proposed to determine
referral and failure criteria for photoscreening. This has
been termed a problem of a “moving goal post.” While
having several sets of referral criteria appears bothersome
from a standpoint of research and standardization, it
clearly is beneficial in some settings. In remote, sparsely
populated areas, it is very expensive to bring a child with a
suspected eye problem in for a formal eye examination. In
Journal of AAPOS
Volume 6 Number 2 April 2002
Alaska, for example, the child often needs to be flown to
the examination location with his or her parents.15 This is
much different than the situation in urban areas, where the
costs associated with a false-positive overreferral are much
less. We were aware of the problems that a low predictive
value positive would bring to the reputation of a volunteer-led program, and therefore developed stringent referral criteria. Other programs having less need to control
false positives may choose a less stringent set of referral
criteria.
Photoscreening will continue to have a role in detecting
amblyogenic factors prior to the onset of literacy. Further
refinements in vision screening technology will likely produce digital cameras and digital analysis systems.16 However, problems with the control of accommodation and the
over-detection of astigmatism will continue to exist. Further research is needed to continue to refine referral criteria, determine the natural history of various levels of
amblyogenic factors, and identify children who need intervention on the basis of their refractive error. It should
also be remembered that the efficacy of photoscreening,
while now quite well established in the ophthalmic literature, is not well documented when compared with traditional vision screening, because most traditional vision
screening techniques have never been adequately validated
in the field. It is only through further investigation that the
knowledge to support a preschool vision screening mandate will develop.
References
1. Simons BD, Siatkowski RM, Schiffman JC, Berry BE, Flynn JT.
Pediatric photoscreening for strabismus and refractive errors in a
high-risk population. Ophthalmology 1999;106:1073-80.
2. Tong PY, Enke-Miyazaki E, Bassin RE, Tielsen JM, Stager DR Sr,
Beauchamp GR, et al. Screening for amblyopia in preverbal children
with photoscreening photographs. National Children’s Eye Care
Foundation Vision Screening Study Group. Ophthalmology 1998;
105:856-63.
3. Ottar WL, Scott WE, Holgado SI. Photoscreening for amblyogenic
factors. J Pediatr Ophthalmol Strabismus 1995;32:289-95.
4. Weinand F, Gräf M, Demming K. Sensitivity of the MTI Photoscreener for amblyogenic factors in infancy and early childhood.
Graefes Arch Clin Exp Ophthalmol 1998;236:801-5.
5. Lewis RC, Marsh-Tootle WL. The reliability of interpretation of
photoscreening results with the MTI PS-100 in Headstart school
children. J Am Optom Assoc 1995;66:429-34.
6. Holgado SI, Arfeli S, Gomez-Demmel E, Espinosa J. Comparative
study of the MTI Photoscreener,™ visual acuity, and Lang stereopsis
test for amblyogenic factors in mentally delayed children. Am Orthoptic J 1998;48:122-30.
7. Tong PY, Bassin RE, Enke-Miyazaki E, Macke JP, Tielsch JM,
Stager DR, et al. Screening for amblyopia in preverbal children with
photoscreening photographs II. Ophthalmology 2000;107:1623-9.
8. Donahue SP, Johnson TM, Leonard-Martin TC. Screening for
amblyogenic factors using a volunteer lay network: initial results
from a statewide program targeting preliterate children. Ophthalmology 2000;107:1637-44.
9. Miller JM, Harvey EM. Spectacle prescribing recommendations of
AAPOS members. J Pediatr Ophthalmol Strabismus 1998;35:51-2.
10. Donahue SP, Johnson TM. Age-based refinement of referral criteria
for photoscreening. Ophthalmology 2001;108:2309-14.
Journal of AAPOS
Volume 6 Number 2 April 2002
Donahue et al
11. Kutschke PJ, Scott WE, Keech RV. Anisometropic amblyopia. Ophthalmology 1991;98:258-63.
12. Tong PY, Macke JP, Bassin RE, Everett M, Enke-Miyazaki E,
Tielsch JM, et al. Screening for amblyopia in preverbal children with
photoscreening photographs III. Ophthalmology 2000;107:1630-6.
13. Atkinson J, Braddick O, Robier B, Anker S, Ehrlich D, King J, et al.
Two infant vision screening programmes: prediction and prevention
of strabismus and amblyopia from photo- and videorefractive screening. Eye 1996;10:189-98.
91
14. The MTI Photoscreener Instruction and Interpretation Manual.
Cedar Falls, IA: Medical Technology, Inc, 1994.
15. Arnold RW, Gionet EG, Jestrzebski AI, Kovtoun TA, Machida CJ,
Armitage MD, et al. The Alaska Blind Child Discovery Project:
rationale, methods and results of 4000 screenings. Alaska Med 2000;
42:58-72.
16. Granet DB, Hoover A, Smith AR, Brown SI, Bartsch DU, Brody B.
A new objective digital computerized vision screening system. J Pediatr Ophthalmol Strabismus 1999;36:251-6.
An Eye on the Arts – The Arts on the Eye
The Song of the Blindman
I am blind, you out there—that is a curse, against one’s will, a contradiction,
a heavy daily burden. I lay my hand on the arm of my wife, my grey hand upon
her greyer grey, as she guides me through empty spaces.
You move about and stir, and imagine your sounds differing from stone to
stone. But you are mistaken; I alone live and suffer and complain, for in me is
an endless crying, and I do not know whether it is my heart that cries or my
bowels.
Do you recognize these songs? You never sang them, not quite with this
intonation. For you every morning brings its new light warm through your
open windows. And you have the feeling from face to face that tempts you to be
indulgent.
—Rainer Maria Rilke (translated by Albert Ernest Flemming)