1040-5488/01/7804-0215/0 VOL. 78, NO. 4, PP.

215–222
OPTOMETRY AND VISION SCIENCE
Copyright © 2001 American Academy of Optometry

ORIGINAL ARTICLE

Retinoscopy in Infants Using a Near

Noncycloplegic Technique, Cycloplegia with
Tropicamide 1%, and Cycloplegia with
Cyclopentolate 1%
J. DANIEL TWELKER, OD, FAAO and DONALD O. MUTTI, OD, PhD, FAAO
University of California, Berkeley, School of Optometry, Berkeley, California (JDT), The Ohio State University College of Optometry,
Columbus, Ohio (DOM)

ABSTRACT: Purpose. This study compares retinoscopy in infants using a near noncycloplegic technique, cycloplegia
with tropicamide 1%, and cycloplegia with cyclopentolate 1%. The study sample included 29 healthy, nonstrabismic
infants 4 to 7 months of age (mean 5.71 months). Methods. Each study subject was examined at two separate visits an
average of 2 weeks apart (mean [ⴞSD] 14 ⴞ 9 days). The examiner completed a case history, iris color grading,
confrontation tests, and noncycloplegic near retinoscopy in a dark room and then instilled a drop of topical anesthetic
in each eye followed by 2 drops of cycloplegic agent separated by 5 min. Retinoscopy was performed 25 to 30 min after
the first drops were instilled. The bottles were masked, and the drop administered at the first visit was randomly
assigned. Results. On a scale of 0 to 4.9, the median iris grade was 4.0, which corresponds to a brown or darkly
pigmented iris. All reported retinoscopy results are for the horizontal meridian of the right eye. The mean refractive
error using noncycloplegic near retinoscopy was ⴙ0.94 D (ⴞ1.19 D). The mean refractive error was ⴙ1.81 D (ⴞ1.19
D) with tropicamide and ⴙ1.88 D (ⴞ1.45 D) with cyclopentolate. There was no statistically or clinically significant
difference between the two cycloplegic measurements using different diagnostic agents (t ⴝ ⴚ0.46, p ⴝ 0.65). The
mean difference between noncycloplegic and cycloplegic retinoscopy was 0.89 D (ⴞ0.66 D) with tropicamide (t ⴝ
ⴚ6.57, p < 0.0001) and 1.04 D (ⴞ0.94 D) with cyclopentolate (t ⴝ ⴚ5.38, p < 0.0001; all two-sided paired t-tests).
There were no serious adverse reactions with either agent, although one infant temporarily developed redder than
normal cheeks after instillation of cyclopentolate. Conclusion. Our results suggest that tropicamide is as effective as
cyclopentolate for the measurement of refractive error in most healthy, nonstrabismic infants. (Optom Vis Sci 2001;78:
215–222)

Key Words: cyclopentolate, cycloplegia, infant, retinoscopy, tropicamide

D
uring the routine examination of nonstrabismic infants,
the eye care practitioner must decide whether or not to
use a cycloplegic agent. A useful noncycloplegic method
is described by Mohindra.1, 2 This procedure is completed at 50
cm in a dark room. The intensity of the light is maintained at a
minimum, and the child is encouraged to fixate the light with one
eye while the other is occluded by the parent. The method is based
on the observation that when viewing a retinoscope light monoc-
ularly at 50 cm in a darkened room, the accommodative response
of adult subjects remains stable at 0.70 D.3 The American Opto-
metric Association recommends the near noncycloplegic tech-
nique when frequent follow-up is necessary, when the child is
extremely anxious about instillation of cycloplegic agents, and
when the child has had or is at risk for an adverse reaction to
cyclopentolate or tropicamide.4
Saunders and Westall5 completed a repeatability study on 74
infants and children, comparing the Mohindra technique with
retinoscopy after the instillation of cyclopentolate. Noncycloplegic
retinoscopy revealed on average 0.39 D less hyperopia than cyclo-
plegic retinoscopy. The authors concluded that noncycloplegic
retinoscopy could be substituted for cycloplegic retinoscopy but
qualified their recommendation by stating “when a poor confi-
dence near retinoscopy result is obtained, a cycloplegic refraction is
advised.” Borghi and Rouse6 also found a good correlation be-

Optometry and Vision Science, Vol. 78, No. 4, April 2001

216 Cycloplegic Retinoscopy in Infants—Twelker & Mutti
tween the near noncycloplegic retinoscopy technique and cyclo-
plegic retinoscopy with cyclopentolate. Cycloplegic retinoscopy
with cyclopentolate found on average 0.50 to 0.75 D more hyper-
opia than noncycloplegic retinoscopy.
Wesson et al.,7 on the other hand, recommended that caution be
used when substituting near noncycloplegic retinoscopy for cyclo-
plegic refraction. In their infant group, they found on average 2.12
D more hyperopia using cycloplegic retinoscopy with cyclopento-
late compared with the near noncycloplegic technique.
To reliably assess refractive error, many clinicians choose to
inhibit accommodation with a cycloplegic agent. Cyclopentolate is
a widely used cycloplegic agent with rapid onset and relatively
short duration. It is commercially available in strengths of 0.5%
and 1.0%. The 1% concentration is often recommended for in-
fants and children ⬍12 years of age.8 Some practitioners recom-
mend the 0.5% concentration for neonates and infants,9, 4 pre-
sumably to decrease the chances of systemic side effects. We
selected the 1% concentration for our study because it is consid-
ered to be the gold standard by which other cycloplegic agents are
compared.
Cyclopentolate provides a depth of cycloplegia that is sufficient
for the majority of cases because it limits residual accommodation
to between 1.00 and 2.50 D.10 –15 Priestly and Medine16 found
that residual accommodation ranged between 0.50 and 1.75 D,
with an average of 1.25 D. The time course is acceptable for clinical
practice, with maximal cycloplegia occurring within 30 to 60 min
of instillation of the first drop. O’Connor-Davies et al.8 found that
the average time to maximum cycloplegia was 34 min, with a range
of 10 to 55 min.
The amount of iris pigmentation has been reported to affect the
time course of cycloplegic agents. Manny et al.17 found that for
individuals with dark irises, 30 to 40 min is required for maximal
cycloplegia with 1 drop of cyclopentolate 1%. The magnitude of
residual accommodation for individuals with dark irises was simi-
lar to that found in individuals with light irises at 10 min. Lova-
sik18 reported that the rate of accommodative loss was slower in
adults with dark irises compared with light irises. The rate of ac-
commodative loss for cyclopentolate 1% after instillation of pro-
paracaine was 1.80 ⫾ 0.24 D/min for blue irises compared with
1.38 ⫾ 0.18 D/min for brown irises.
Some clinicians, however, avoid the use of cyclopentolate for
routine eye examination in infants because of the concern for po-
tential systemic side effects. Over the years, numerous case reports
have been published which associate the drug with central nervous
system side effects such as cerebellar dysfunction, visual and tactile
hallucinations, drowsiness, ataxia, seizures, facial flushing, and
tachycardia.19 –22 Many of these adverse reactions occurred with
cyclopentolate 2%, whereas most practitioners today use cyclopen-
tolate 1%. Nevertheless, because of these reports, some eye care
practitioners reserve cyclopentolate for the examination of infants
and children with strabismus or other ocular anomalies.
Tropicamide is another widely used cycloplegic agent. It is com-
mercially available in concentrations of 0.5% and 1.0%. For the
purposes of cycloplegia, the 1% concentration is recommended
over the 0.5%.8 Using 2 drops of tropicamide 1%, residual accom-
modation is usually ⬍2.00 D and the interval of maximal cyclo-
plegia is 20 to 35 min.12 The duration of cycloplegia is short-lived,
however, and quickly recovers after 35 min accommodation.
Optometry and Vision Science, Vol. 78, No. 4, April 2001
Mishima23 and Maurice and Mishima24 estimated that tropicam-
ide is eliminated from the receptor sites about three times more
quickly than cyclopentolate, thereby significantly reducing its bio-
availability for inducing cycloplegia. The amount of iris pigmen-
tation affects tropicamide as well. The rate of accommodative loss
for tropicamide 1% after instillation of proparacaine was 1.26 ⫾
0.12 D/min for blue irises compared with 0.72 ⫾ 0.24 D/min for
brown irises. Tropicamide is a safe diagnostic pharmaceutical
agent. In a review of 15,000 diagnostic pharmaceutical applica-
tions, Yolton et al.25 reported no adverse systemic effects for
tropicamide.
The main drawback of tropicamide is that some studies have
demonstrated a lack of deep cycloplegic effect. Milder13 found that
six cases in his series up to 9 years of age had an average of 6.25 D
of residual accommodation after 30 min. In 20 cases from ages 10
to 14 years, residual accommodation was an average of 3.65 D.
Lovasik18 estimates that 1 drop of tropicamide 0.5% or 1.0%
might leave 28% to 40% of the normal accommodative amplitude
intact. He concludes that this amount of residual accommodation
is unsatisfactory for most refractive purposes. Hiatt and Jerkins26
found that tropicamide was less effective as a cycloplegic for pre-
school children with esotropia.
The studies of residual accommodation are important and can
be a helpful guide to the usefulness of a cycloplegic agent. Never-
theless, the practitioner is generally most interested in a reliable
measure of refractive error. Despite its poorer performance as a
cycloplegic agent compared with cyclopentolate in terms of inhib-
iting accommodation, tropicamide has been found to provide ad-
equate cycloplegia for measurement of refractive error. Egashira et
al.27 reported that in hyperopic 6 to 12-year-old children, there
was no statistically significant difference between cyclopentolate
and tropicamide for either cycloplegic retinoscopy or subjective
refraction. There was, however, a statistically significant but clin-
ically unimportant bias of 0.14 D toward more hyperopia with
cyclopentolate when autorefraction was used.
We wanted to evaluate whether tropicamide might be a useful
cycloplegic agent for the routine examination of nonstrabismic
infants, taking advantage of its wider margin of safety and faster
recovery time compared with cyclopentolate, as well as its potential
for revealing more hyperopia than noncycloplegic retinoscopy.
METHODS
Twenty-nine subjects, 18 girls and 11 boys, aged 4 to 7 months
participated in the study (mean age [⫾SD] 5.7 ⫾ 0.7 months)
after appropriate informed consent procedures administered to the
parents in a language in which they were fluent. The study was
performed at La Clinica de la Raza in Oakland, California, and all
of the participants were Hispanic, which corresponds to the patient
base of the Clinic. The protocol and informed consent forms were
reviewed and approved by the University of California, Berkeley
Committee for the Protection of Human Subjects. Each study
subject was examined at two separate visits an average of 2 weeks
apart (mean 14 ⫾ 9 days). All subjects had normal ocular health
with no strabismus.
All of the examination procedures in this study were completed
by one examiner (JDT). He completed a case history and confron-
tation tests including Hirschberg angles, near point of conver-
 Cycloplegic Retinoscopy in Infants—Twelker & Mutti 217

gence, pupil response testing, and extraocular muscle motility. The

examiner graded iris color using a slight modification of the
method described by Seddon et al.28 and recorded a grade of 0 to
4.9 with decimalization to the nearest 0.1 units. The four standard
plates were assigned grades 1 through 4, where standard A was the
least pigmented (grade 1) and standard D the most heavily pig-
mented (grade 4).
During all retinoscopy procedures, individual trial lenses rather
than lens bars were used to neutralize the reflex. The examiner first
completed noncycloplegic near retinoscopy in a dark room as de-
scribed by Mohindra.2 Rather than the 1.25 D correction recom-
mended by Mohindra, we subtracted 0.75 D from the near reti-
noscopy result as recommended by Saunders and Westall.5 The
Saunders and Westall study used a larger sample size than the
Mohindra study, and more important, the conclusions were based
on a sample of infant subjects rather than adult subjects.
The examiner inadvertently used a nonstandard technique for
the first five study subjects. Instead of covering the nontested eye,
the examiner performed retinoscopy while the infant was binocu-
lar. The first study subject received binocular testing at both visits. FIGURE 1.
Subjects 2 through 5 received binocular testing on the first visit Difference vs. mean graph for iris grade. The x axis shows the mean of the
only and the proper monocular technique at the second visit. Sub- visit 1 and visit 2 iris grades. The y axis shows the visit 1 minus the visit
jects 6 through 29 received the monocular technique at both visits. 2 grades. The solid line represents the mean difference between the two
grades for all subjects. There is no statistically significant bias between the
After noncycloplegic retinoscopy, the examiner instilled a drop visit 1 and visit 2 grades. The gray shaded area represents the 95% limits
of proparacaine hydrochloride 0.5% (AK-Taine, Akorn, Buffalo of agreement between visit 1 and visit 2.
Grove, IL) in each eye followed by 2 drops of tropicamide 1% or
cyclopentolate 1%, separated by 5 min (Tropicacyl, Akorn, Buf-
falo Grove, IL or AK-Pentolate, Akorn, Abita Springs, LA). The subjects had a brown or dark brown iris. Iris color was not analyzed
examiner then waited another 20 min. In all cases, the subject and further due to this limited variation.
parent were called into the examination room 25 min after the
initial eye drop. Depending on how long it took the subject and
parent to settle into the examination chair, retinoscopy was per-
Noncycloplegic Retinoscopy
formed 25 to 30 min after the initial eye drop. The same time For the purposes of assessing repeatability of the spherical com-
course was used for both cycloplegic agents. After retinoscopy, the ponent of refractive error, the reported results are for the horizontal
examiner completed a dilated fundus examination. meridian of the right eye. Study subjects 1 through 5 were excluded
The bottles containing both cycloplegic agents were masked, from the analysis of noncycloplegic results because of nonstandard
and the drop that was administered at the first visit was randomly technique as described in the methods section.
assigned. At the second visit, the examiner was masked to all prior The mean noncycloplegic retinoscopy was ⫹0.84 D (⫾1.22 D)
noncycloplegic and cycloplegic retinoscopy findings. Some work- for visit 1 and ⫹1.04 D (⫾1.31 D) for visit 2. The mean difference
ers have noticed that there is a difference in odor between tropic- between visits was ⫺0.20 D (⫾0.86 D). Fig. 2 shows the difference
amide 1% and cyclopentolate 1%. The examiner was unaware of between visits on the y axis vs. the mean of noncycloplegic retinos-
this difference at the time of the examinations and was not un- copy on the x axis. The solid line represents the mean difference
masked by any difference in odor at the arm’s length distance used between visits of ⫺0.20 D. The gray shaded area represents the
for drop instillation during testing. 95% limits of agreement of ⫾1.78 D between visits. We per-
formed a linear regression to determine whether there was a range
effect between the difference (visit 1 minus visit 2) and mean
RESULTS refractive error. The analysis suggests that there is no relationship
between retinoscopic difference and mean refractive error (y ⫽
Iris Grading
⫺0.08x ⫹ 0.12, R2 ⫽ 0.01, p ⫽ 0.60). The mean of both noncy-
The median iris grade for visit 1 was 4.0 (range 1.4 to 4.9) and cloplegic visits was ⫹0.94 D (⫾1.19 D).
for visit 2 was 4.1 (range 0.8 to 4.8), which corresponds to a brown
or darkly pigmented iris. There was no statistically significant dif-
Cycloplegic Retinoscopy
ference between grades for visits 1 and 2 (t ⫽ ⫺1.00, p ⫽ 0.33,
two-sided paired t-test). Fig. 1 shows the difference between visits Table 1 presents the refractive data for the horizontal meridian
on the y axis vs. the mean of iris grade on the x axis. The solid line of the right eye under noncycloplegic, tropicamide, and cyclopen-
represents the mean difference between visits of ⫺0.1 units tolate conditions. The mean retinoscopy results were ⫹1.81 D
(⫾0.3). The gray shaded area represents the 95% limits of agree- (⫾1.19 D) for tropicamide 1% and ⫹1.88 D (⫾1.45 D) for
ment of ⫾0.61 units between visits. As seen in Fig. 1, nearly all cyclopentolate 1%. The mean difference between the two cyclo-

Optometry and Vision Science, Vol. 78, No. 4, April 2001

218 Cycloplegic Retinoscopy in Infants—Twelker & Mutti
FIGURE 2.
Difference vs. mean graph for noncycloplegic retinoscopy. The x axis
shows the mean of visit 1 and visit 2 retinoscopy. The y axis shows the
difference between visits. The solid line represents the mean difference
between visits of 0.20 D. The gray shaded area represents the 95% limits
of agreement of ⫾1.78 D.
plegic agents was 0.07 D (⫾0.80 D). There was no statistically or
clinically significant difference between the two measurements us-
ing different cycloplegic agents (t ⫽ ⫺0.46, p ⫽ 0.65, two-sided
paired t-test). Fig. 3 shows the difference between the two cyclo-
plegic agents on the y axis vs. the mean of both on the x axis. The
solid line represents the mean difference between visits of 0.07 D.
The gray shaded area represents the 95% limits of agreement of
⫾1.65 D between visits. There were no serious adverse reactions
with either agent, although one infant temporarily developed red-
der than normal cheeks after instillation of cyclopentolate.
We performed a linear regression on the relative difference be-
tween cycloplegic measurements (cyclopentolate minus tropicam-
ide) and the mean refractive error as shown in Fig. 3. The purpose
was to investigate any linear relationship that might be present
between the two variables. The results suggest that the retinoscopic
difference between agents increases with increasing mean refractive
error, although it is not significant at the 95% level (y ⫽ 0.22x ⫺
0.33, R2 ⫽ 0.12, p ⫽ 0.07).
Difference between Noncycloplegic and
Cycloplegic Retinoscopy
Table 1 shows that the mean difference between noncycloplegic
and cycloplegic retinoscopy was 0.89 D (⫾0.66 D; t ⫽ ⫺6.57,
p ⬍ 0.0001) with tropicamide and 1.04 D (⫾0.94 D; t ⫽ ⫺5.38,
p ⬍ 0.0001; all two-sided paired t-tests) with cyclopentolate.
Fig. 4a shows the difference between the tropicamide and non-
cycloplegic retinoscopy on the y axis vs. the mean of both on the x
axis. The solid line represents the mean difference between mea-
surements of 0.89 D. The gray shaded area represents the 95%
limits of agreement of ⫾1.37 D between visits. Fig. 4b shows the
difference between the cyclopentolate and noncycloplegic retinos-
Optometry and Vision Science, Vol. 78, No. 4, April 2001
copy on the y axis vs. the mean of both on the x axis. The solid line
represents the mean difference between measurements of 1.04 D.
The gray shaded area represents the 95% limits of agreement of
⫾1.95 D between visits.
We performed a linear regression to determine whether there
was a range effect between the difference (cycloplegic minus non-
cycloplegic) and mean refractive error. The analysis suggests that
there is no relationship between the difference and mean refractive
error for tropicamide or cyclopentolate (tropicamide, y ⫽ 0.03x ⫹
0.84, R2 ⬍ 0.01, p ⫽ 0.78; cyclopentolate, y ⫽ 0.17x ⫹ 0.79, R2
⬍ 0.01, p ⫽ 0.31).
Astigmatism
Table 2 presents the astigmatic refractive data under noncy-
cloplegic (mean of both visits), tropicamide, and cyclopentolate
conditions where the horizontal meridian was subtracted from the
vertical meridian. Therefore, negative astigmatic retinoscopy re-
sults indicate with-the-rule astigmatism. The mean retinoscopy
was ⫺1.38 D (⫾1.00 D) for the noncycloplegic technique, ⫺1.80
D (⫾1.32 D) for tropicamide, and ⫺1.81 D (⫾1.06 D) for cyclo-
pentolate. The difference between conditions is also presented.
The mean difference was ⫺0.49 D (⫾0.96 D; t ⫽ ⫺2.50, p ⫽
0.02) using tropicamide (tropicamide minus noncycloplegic),
⫺0.52 D (⫾0.82 D; t ⫽ ⫺3.11, p ⫽ 0.0049; all two-sided paired
t-tests) for cyclopentolate (cyclopentolate minus noncycloplegic),
and ⫺0.01 D (⫾0.92 D) between cycloplegics (cyclopentolate minus
tropicamide). There was no statistically or clinically significant differ-
ence between the two astigmatism measurements using different cy-
cloplegic agents (t ⫽ 0.05, p ⫽ 0.96, two-sided paired t-test).
DISCUSSION
Tropicamide 1% and cyclopentolate 1% resulted in similar ret-
inoscopy results for most infant subjects. The mean difference was
0.07 D (⫾0.80 D, t ⫽ ⫺0.46, p ⫽ 0.65, two-sided paired t-test)
for the horizontal meridian of the right eye. This result suggests
that tropicamide is as effective when measuring refractive error as
cyclopentolate in a nonstrabismic infant population. The dark iris
color of the sample makes this a particularly interesting compari-
son because both cyclopentolate and tropicamide have been re-
ported to be less effective in persons with dark irises compared with
light irises.
It might be argued that our subject sample was even more hy-
peropic than was measured with cyclopentolate. Maximum cyclo-
plegia ranges between 20 to 30 min for tropicamide12 and 30 to 40
min for cyclopentolate.17 We performed retinoscopy 25 to 30 min
after the first drop was instilled. Therefore, for tropicamide cyclo-
plegia, the examiner performed retinoscopy in the range of maxi-
mum cycloplegia. This should not bias our results substantially,
however; the differences reported by Manny et al.17 in optometer-
measured residual accommodation 20 and 30 min after cyclopen-
tolate 1% in subjects with dark irises were small (about 0.25 D).
When cycloplegia is incomplete, the examiner sometimes notes
that the subject is accommodating during retinoscopy. This is seen
as a fluctuation in the retinoscopy reflex and/or the refractive error
measurement. Although there was no formal evaluation of this in
the testing protocol, the examiner did not note any cases of incom-
 Cycloplegic Retinoscopy in Infants—Twelker & Mutti 219

TABLE 1.
Refraction data by subject in the horizontal meridian for noncycloplegic and cycloplegic retinoscopy after tropicamide
1% and cyclopentolate 1%.

Retinoscopy (D) Difference (D)

Subject Noncycloplegic Tropicamide Cyclopentolate Cyclopentolate
No. (mean of visit 1 Tropicamide Cyclopentolate minus minus minus
and visit 2) Noncycloplegic Noncycloplegic Tropicamide

1 — ⫹1.00 ⫹2.00 — — ⫹1.00

2 — ⫹2.00 ⫹2.50 — — ⫹0.50
3 — ⫹1.00 ⫺1.00 — — ⫺2.00
4 — ⫹1.00 0.00 — — ⫺1.00
5 — ⫹3.50 ⫹3.50 — — 0.00
6 ⫹2.75 ⫹3.00 ⫹3.50 ⫹0.25 ⫹0.75 ⫹0.50
7 ⫹0.25 ⫹1.50 ⫹2.25 ⫹1.25 ⫹2.00 ⫹0.75
8 ⫹0.75 ⫹1.25 ⫹0.75 ⫹0.50 0.00 ⫺0.50
9 ⫹0.50 ⫹0.50 ⫹0.50 ⫹0.00 0.00 0.00
10 ⫹2.75 ⫹3.25 ⫹3.50 ⫹0.50 ⫹0.75 ⫹0.25
11 ⫹1.50 ⫹1.00 ⫹1.50 ⫺0.50 0.00 ⫹0.50
12 ⫹2.00 ⫹3.50 ⫹3.75 ⫹1.50 ⫹1.75 ⫹0.25
13 ⫺0.50 ⫹1.25 ⫹1.50 ⫹1.75 ⫹2.00 ⫹0.25
14 ⫹4.25 ⫹5.50 ⫹5.50 ⫹1.25 ⫹1.25 0.00
15 ⫹0.50 ⫹1.75 ⫹1.50 ⫹1.25 ⫹1.00 ⫺0.25
16 ⫹0.50 ⫹1.00 ⫹1.25 ⫹0.50 ⫹0.75 ⫹0.25
17 ⫹1.25 ⫹1.50 ⫺0.50 ⫹0.25 ⫺1.75 ⫺2.00
18 ⫹1.25 ⫹2.50 ⫹3.25 ⫹1.25 ⫹2.00 ⫹0.75
19 ⫹0.50 ⫹1.25 ⫹2.50 ⫹0.75 ⫹2.00 ⫹1.25
20 ⫺0.25 ⫹1.25 ⫹1.25 ⫹1.50 ⫹1.50 0.00
21 ⫺1.00 ⫺0.25 ⫹1.25 ⫹0.75 ⫹2.25 ⫹1.50
22 ⫺0.25 ⫹1.00 ⫹0.50 ⫹1.25 ⫹0.75 ⫺0.50
23 ⫹1.50 ⫹1.25 ⫹1.50 ⫺0.25 0.00 ⫹0.25
24 ⫹0.25 ⫹1.50 ⫹1.50 ⫹1.25 ⫹1.25 0.00
25 ⫺0.63 ⫹1.00 ⫹0.25 ⫹1.63 ⫹0.88 ⫺0.75
26 ⫹0.88 ⫹1.00 ⫹1.50 ⫹0.13 ⫹0.63 ⫹0.50
27 ⫹1.13 ⫹2.75 ⫹2.25 ⫹1.63 ⫹1.13 ⫺0.50
28 ⫹1.25 ⫹2.50 ⫹3.25 ⫹1.25 ⫹2.00 ⫹0.75
29 ⫹1.50 ⫹3.25 ⫹3.50 ⫹1.75 ⫹2.00 ⫹0.25

Mean ⫹0.94 ⫹1.81 ⫹1.88 ⫹0.89a ⫹1.04a ⫹0.07

SD 1.19 1.19 1.45 0.66 0.94 0.80
a
p ⬍ 0.0001; two-sided paired t-test.

plete cycloplegia during the infant examinations. Another poten-
tial sign of incomplete cycloplegia is an increased astigmatism mea-
surement. As the examiner moves from the horizontal meridian to
the vertical meridian, a patient for whom cycloplegia is not com-
plete might accommodate, resulting in increased astigmatic mea-
sures. Table 2 shows that the difference between tropicamide and
cyclopentolate astigmatic refractive error for the right eye was not
significant with an average difference of ⫺0.01 D (⫾0.92 D; t ⫽
0.05, p ⫽ 0.96, two-sided paired t-test).
Both agents clearly had a cycloplegic effect. Both tropicamide
and cyclopentolate uncovered more hyperopia than noncyclople-
gic retinoscopy (0.89 ⫾ 0.66 D, t ⫽ ⫺6.57, p ⬍ 0.0001 and 1.04
⫾ 0.94 D, t ⫽ ⫺5.38, p ⬍ 0.0001, respectively). Furthermore,
cycloplegic retinoscopy using tropicamide (and cyclopentolate)
was much easier to perform compared with noncycloplegic reti-
noscopy. The pupils were larger, which made the motion of the
reflex much easier to interpret. More important, the infant subjects
sat much more quietly and maintained fixation much better than
when we used the noncycloplegic technique as described by Mo-
hindra.2 When the parent gently covered the eye not being tested,
most infants tried to avoid the cover and often became agitated or
began to cry. This made retinoscopy more difficult to perform.
Under binocular conditions, on the other hand, most infants were
relatively cooperative during retinoscopy.
The process of instilling eye drops also causes some temporary
agitation and crying in most infants. But, unlike the near Mohin-
dra technique, the agitation occurs 25 min earlier than the retinos-
copy procedure, not simultaneously. Most infants recover very
quickly from any agitation induced by instilling eye drops.
We found that the linear relationship between the difference
between cycloplegic measurements (cyclopentolate minus tropic-
amide) and the mean refractive error was not statistically signifi-
cant (y ⫽ 0.22x ⫺ 0.33, R2 ⫽ 0.12, p ⫽ 0.07). Because of the
positive sign on the slope, a type II error here suggests that retinos-

Optometry and Vision Science, Vol. 78, No. 4, April 2001

220 Cycloplegic Retinoscopy in Infants—Twelker & Mutti

FIGURE 3.
Difference vs. mean graphs for cycloplegic retinoscopy. The x axis shows
the mean of cyclopentolate and tropicamide retinoscopy. The y axis
shows the difference between both agents. The solid line represents the
mean difference between visits of 0.07 D. The gray shaded area represents
the 95% limits of agreement of ⫾1.65 D.

copy using tropicamide cycloplegia might underestimate refractive

error in cases of moderate to high hyperopia. Closer examination of
Fig. 3 indicates that this is not the case. The trend toward a positive
slope appears to be influenced by two points that indicate a more
hyperopic result with tropicamide, not cyclopentolate, at near em-
metropic mean refractive errors. When refractive error is more
hyperopic than ⫹2.00 D, the two cycloplegic agents give results
that are within ⫾1.00 D of each other with no evidence of under-
estimation of hyperopia with tropicamide.
Although this study is not intended to be a cross-sectional study
of refractive error in 6-month-old infants, it is interesting to note
that both tropicamide and cyclopentolate measures of refractive
error (⫹1.81 ⫾ 1.19 D and ⫹1.88 ⫾ 1.45 D, respectively) are
similar to those found in large studies of refractive error in infants.
More hyperopia was found by Ehrlich and co-workers29 (an aver-
age mean for the right eye most positive meridian of ⫹2.50 ⫾ 1.40
D in 254 infants) and slightly less hyperopia by Frane and co-
workers30 (an average of ⫹1.31 ⫾ 1.10 D in 216 infants) in two
separate studies of 9-month-old infants using cyclopentolate cyclo-
plegic retinoscopy. Our results might be applicable to a broad
range of subjects of various ethnicities and iris colors.
When interpreting the results of any research investigation, one
must consider the potential for bias due to measurement error.
This project was completed entirely by one examiner, therefore
there was no interobserver bias. We can assess intraobserver repeat-
ability for noncycloplegic retinoscopy because the same procedure
was completed at both examination visits. The 95% limits of agree-
ment were ⫾1.78 D. In this study, it is not possible to directly
assess intraobserver repeatability for cycloplegic retinoscopy be-
cause a different cycloplegic agent was used at each study visit. If
you make the assumption that there was little to no variability due
to using a different cycloplegic agent from visit to visit, which this
FIGURE 4.
Difference vs. mean graphs for cycloplegic vs. noncycloplegic retinos-
copy: (a) tropicamide vs. noncycloplegic and (b) cyclopentolate vs. non-
cycloplegic retinoscopy. The x axis shows the mean for cycloplegic and
noncycloplegic retinoscopy. The y axis shows the difference between
both conditions. The solid line represents the mean difference between
visits of (a) 0.89 D for tropicamide and (b) 1.04 D for cyclopentolate. The
gray shaded area represents the 95% limits of agreement of (a) ⫾1.37 D
for tropicamide and (b) ⫾1.95 D for cyclopentolate.
study supports, then the 95% limits of agreement would be ⫾1.65
D. In a study comparing retinoscopy in a sample of 20 hyperopic
6- to 12-year-old children,27 the 95% limits of agreement were
⫾0.63 D. This implies that retinoscopy is more difficult to per-
form in infants compared with school-age children, which is not
unexpected. A second consideration when no significant differ-
ences are found between two techniques is the adequacy of the
sample size, or the power of the study to find a difference if one
exists. The study power is ⬎90% to detect a difference of 0.50 D
between cycloplegic agents at a sample size of 29 and standard
deviation of differences of ⫾0.80 D.
Most infants tend to hold a steady gaze for only short periods of
time. Therefore, the examiner must be quick in obtaining an end
point, while also trying to maximize accuracy. Even though fixa-

Optometry and Vision Science, Vol. 78, No. 4, April 2001

 Cycloplegic Retinoscopy in Infants—Twelker & Mutti 221

TABLE 2.
Astigmatism data by subject (vertical minus horizontal meridian) for noncycloplegic and cycloplegic retinoscopy
following tropicamide 1% and cyclopentolate 1%.

Retinoscopy (D) Difference (D)

Subject Noncycloplegic Tropicamide Cyclopentolate Cyclopentolate
No. (mean of visit 1 Tropicamide Cyclopentolate minus minus minus
and visit 2) Noncycloplegic Noncycloplegic Tropicamide

1 — ⫺1.00 ⫺2.00 — — ⫺1.00

2 — ⫺0.50 ⫺1.00 — — ⫺0.50
3 — ⫺4.00 ⫺0.50 — — ⫹3.50
4 — 0.00 ⫺1.00 — — ⫺1.00
5 — ⫺2.00 ⫺2.50 — — ⫺0.50
6 ⫺1.25 ⫺1.00 ⫺1.00 ⫹0.25 ⫹0.25 ⫹0.00
7 ⫺1.25 ⫺2.00 ⫺2.25 ⫺0.75 ⫺1.00 ⫺0.25
8 ⫺1.50 ⫺3.25 ⫺2.50 ⫺1.75 ⫺1.00 ⫹0.75
9 ⫺1.75 ⫺2.50 ⫺2.75 ⫺0.75 ⫺1.00 ⫺0.25
10 ⫺0.88 ⫺1.50 ⫺1.00 ⫺0.63 ⫺0.13 ⫹0.50
11 ⫺0.75 ⫺0.50 ⫺0.75 ⫹0.25 0.00 ⫺0.25
12 ⫺1.00 ⫺1.50 ⫺2.75 ⫺0.50 ⫺1.75 ⫺1.25
13 ⫺0.75 ⫺0.75 ⫺1.00 0.00 ⫺0.25 ⫺0.25
14 ⫺5.00 ⫺5.00 ⫺5.00 0.00 0.00 0.00
15 ⫺0.75 ⫺0.50 ⫺1.75 ⫹0.25 ⫺1.00 ⫺1.25
16 ⫺1.75 ⫺3.00 ⫺2.50 ⫺1.25 ⫺0.75 ⫹0.50
17 ⫺1.75 ⫺0.75 ⫺1.00 ⫹1.00 ⫹0.75 ⫺0.25
18 ⫺1.38 ⫺1.00 ⫺1.25 ⫹0.38 ⫹0.13 ⫺0.25
19 ⫺0.75 ⫺1.25 ⫺1.75 ⫺0.50 ⫺1.00 ⫺0.50
20 ⫺0.25 ⫺3.25 ⫺3.25 ⫺3.00 ⫺3.00 0.00
21 ⫹0.25 ⫺0.25 0.00 ⫺0.50 ⫺0.25 ⫹0.25
22 ⫺1.00 ⫺0.75 ⫺1.25 ⫹0.25 ⫺0.25 ⫺0.50
23 ⫺0.75 ⫺1.75 ⫺1.00 ⫺1.00 ⫺0.25 ⫹0.75
24 ⫺1.50 ⫺3.25 ⫺3.00 ⫺1.75 ⫺1.50 ⫹0.25
25 ⫺1.00 ⫺2.25 ⫺1.75 ⫺1.25 ⫺0.75 ⫹0.50
26 ⫺2.25 ⫺2.25 ⫺2.50 0.00 ⫺0.25 ⫺0.25
27 ⫺1.88 ⫺1.00 ⫺1.75 ⫹0.88 ⫹0.13 ⫺0.75
28 ⫺1.38 ⫺1.00 ⫺0.75 ⫹0.38 ⫹0.63 ⫹0.25
29 ⫺2.75 ⫺4.50 ⫺3.00 ⫺1.75 ⫺0.25 ⫹1.50

Mean ⫺1.38 ⫺1.80 ⫺1.81 ⫺0.49a ⫺0.52b ⫺0.01

SD 1.00 1.32 1.06 0.96 0.82 0.92
a
p ⫽ 0.02; two-sided paired t-test.
b
p ⫽ 0.0049; two-sided paired t-test.

tion aids can help the infant maintain a steady gaze, the examiner
must be able to identify when he or she is off-axis with respect to
the line of sight, which can bias results. The infant often tries to
grab the individual trial lens that is used for neutralizing the reti-
noscopic reflex, and the lens bar can be even more difficult because
it presents and even larger target. Some infants, for any number of
reasons, simply do not want to sit still or open their eyes. And, of
course, some will cry.
Retinoscopy in infants is inherently imprecise, which can lead to
an inaccurate result. Given only one estimate, it is impossible for
the practitioner to know whether the result is inaccurate. It is
important to think of one estimate, not as the infant’s refractive
error, but rather as one piece of evidence leading to knowing the
infant’s refractive error. Repeated retinoscopy, either during the
same session a few minutes later or on another day, can lead to
increased confidence in the result. Conversely, it can lead to doubt
in the initial retinoscopy result, which is equally important
information.
Noncycloplegic retinoscopy as described by Mohindra,2 al-
though it is simply a modification of standard retinoscopy, takes
time and practice to perform reliably. The authors recommend
practical experience with about 8 to 10 infant patients before the
clinician is comfortable with the procedure. This is a significant
drawback for the primary care optometrist who only infrequently
examines an infant patient. Cycloplegic retinoscopy, on the other
hand, is a procedure that is much more commonly performed in
most clinics. Although it is true that performing the procedure on
a less than cooperative infant can be difficult, the procedure is still
a familiar one. For the clinician who completes a limited number of
infant examinations per year, we recommend cycloplegic retinos-
copy. Specifically, we recommend retinoscopy using tropicamide
because it appears to be effective as cyclopentolate. Its wider mar-

Optometry and Vision Science, Vol. 78, No. 4, April 2001

222 Cycloplegic Retinoscopy in Infants—Twelker & Mutti
gin of safety and faster recovery time are important practical
considerations.
10.
CONCLUSIONS 11.
The American Optometric Association recommends a complete 12.
eye examination at about 6 months of age. This recommendation
could result in the detection of visual abnormalities at an early age, 13.
often when they are more amenable to treatment to prevent am-
blyopia. Good examples of conditions amenable to early treatment 14.
to prevent amblyopia are strabismus, astigmatism, high hyperopia,
and anisometropia. Nevertheless, the examination of infants pre- 15.
sents some important challenges. Certainly, near noncycloplegic
16.
retinoscopy as described by Mohindra is an important technique
that optometrists have at their disposal, but it has been reported to 17.
be unreliable. Cycloplegic retinoscopy using cyclopentolate is the
gold standard by which other methods are compared. However,
some practitioners hesitate to use cyclopentolate unless it is neces-
sary because of reports of rare adverse reactions. This study presents 18.
tropicamide as a viable option for inducing cycloplegia in infants.
Tropicamide appears to be as effective as cyclopentolate for the 19.
measurement of refractive error in most nonstrabismic infants.
20.
ACKNOWLEDGMENTS
21.
This project was supported by National Eye Institute grant K23-EY00372 and
an Ezell Fellowship from the American Optometric Foundation. We thank the
22.
staff of La Clinica de la Raza for their support, especially Adriana Aceves for
her assistance with the recruitment of study participants.
Presented at the American Academy of Optometry Annual Meeting on
December 10, 2000, at Orlando, Florida. 23.
Received October 25, 2000; revision received January 9, 2001.
24.
REFERENCES
25.
1. Mohindra I. A technique for infant vision examination. Am J Optom
Physiol Opt 1975;52:867–70.
2. Mohindra I. A non-cycloplegic refraction technique for infants and
26.
young children. J Am Optom Assoc 1977;48:518–23.
3. Owens DA, Mohindra I, Held R. The effectiveness of a retinoscope
27.
beam as an accommodative stimulus. Invest Ophthalmol Vis Sci
1980;19:942–9.
4. Scheiman MM, Amos CS, Ciner EB, Marsh-Tootle W, Moore BD,
28.
Rouse MW. Pediatric Eye and Vision Examination. Reference Guide
for Clinicians, 2nd ed. Optometric Clinical Practice Guidelines. St.
Louis: American Optometric Association, 1997.
5. Saunders KJ, Westall CA. Comparison between near retinoscopy and
29.
cycloplegic retinoscopy in the refraction of infants and children. Op-
tom Vis Sci 1992;69:615–22.
6. Borghi RA, Rouse MW. Comparison of refraction obtained by “near
retinoscopy” and retinoscopy under cycloplegia. Am J Optom
30.
Physiol Opt 1985;62:169–72.
7. Wesson MD, Mann KR, Bray NW. A comparison of cycloplegic
refraction to the near retinoscopy technique for refractive error deter-
mination. J Am Optom Assoc 1990;61:680–4.
8. O’Connor-Davies PH, Hopkins GA, Pearson RM. The Actions and
Uses of Ophthalmic Drugs, 3rd ed. London: Butterworths, 1989:
96–100.
9. Amos JF, Amos DM. Cycloplegic refraction. In: Bartlett JD, Jaanus
Optometry and Vision Science, Vol. 78, No. 4, April 2001
SD, eds. Clinical Ocular Pharmacology, 3rd ed. Boston: Butter-
worth-Heinemann, 1995:506–7.
Gettes BC, Leopold IH. Evaluation of five new cycloplegic drugs.
Arch Ophthalmol 1953;49:24–7.
Gettes BC. Three new cycloplegic drugs. Arch Ophthalmol 1954;51:
467–72.
Gettes BC, Belmont O. Tropicamide: comparative cycloplegic ef-
fects. Arch Ophthalmol 1961;66:336–40.
Milder B. Tropicamide as a cycloplegic agent. Arch Ophthalmol
1961;66:70–2.
Milder B. An evaluation of Cyclogyl (Compound 75 GT). Am J
Ophthalmol 1953;36:1724–6.
Stolzar IH. A new group of cycloplegic drugs: further studies. Am J
Ophthalmol 1953;36:110–2.
Priestley BS, Medine MM. A new mydriatic and cycloplegic drug.
Am J Ophthalmol 1951;34:572.
Manny RE, Fern KD, Zervas HJ, Cline GE, Scott SK, White JM,
Pass AF. 1% Cyclopentolate hydrochloride: another look at the time
course of cycloplegia using an objective measure of the accommoda-
tive response. Optom Vis Sci 1993;70:651–65.
Lovasik JV. Pharmacokinetics of topically applied cyclopentolate
HCl and tropicamide. Am J Optom Physiol Opt 1986;63:787–803.
Jones LW, Hodes DT. Possible allergic reactions to cyclopentolate
hydrochloride: case reports with literature review of uses and adverse
reactions. Ophthalmic Physiol Opt 1991;11:16–21.
Simcoe CW. Cyclopentolate (Cyclogyl) toxicity. Arch Ophthalmol
1962;67:406–8.
Awan KJ. Adverse systemic reactions of topical cyclopentolate hydro-
chloride. Ann Ophthalmol 1976;8:695–8.
Binkhorst RD, Weinstein GW, Baretz R, Clahane A. Psychotic reac-
tion induced by cyclopentolate (Cyclogyl). Am J Ophthalmol 1963;
55:1243–5.
Mishima S. Clinical pharmacokinetics of the eye. Proctor lecture.
Invest Ophthalmol Vis Sci 1981;21:504–41.
Maurice DM, Mishima S. Ocular pharmacokinetics. In: Sears ML,
ed. Handbook of Experimental Pharmacology, Vol. 69. Berlin:
Springer-Verlag, 1984:19–116.
Yolton DP, Kandel JS, Yolton RL. Diagnostic pharmaceutical agents:
side effects encountered in a study of 15,000 applications. J Am
Optom Assoc 1980;51:113–8.
Hiatt RL, Jerkins G. Comparison of atropine and tropicamide in
esotropia. Ann Ophthalmol 1983;15:341–3.
Egashira SM, Kish LL, Twelker JD, Mutti DO, Zadnik K, Adams
AJ. Comparison of cyclopentolate versus tropicamide cycloplegia in
children. Optom Vis Sci 1993;70:1019–26.
Seddon JM, Sahagian CR, Glynn RJ, Sperduto RD, Gragoudas ES.
Evaluation of an iris color classification system. The Eye Disorders
Case-Control Study Group. Invest Ophthalmol Vis Sci 1990;31:
1592–8.
Ehrlich DL, Braddick OJ, Atkinson J, Anker S, Weeks F, Hartley T,
Wade J, Rudenski A. Infant emmetropization: longitudinal changes
in refraction components from nine to twenty months of age. Optom
Vis Sci 1997;74:822–43.
Frane SL, Sholtz RI, Lin WK, Mutti DO. Ocular components before
and after acquired, nonaccommodative esotropia. Optom Vis Sci
2000;77:633–6.
J. Daniel Twelker
School of Optometry
Berkeley, California 94720-2020
e-mail: twelker@uclink4.berkeley.edu