A comparison of cycloplegic
refraction to the near retinoscopy
MICHAEL D. WESSON, 0 .0 ., M.S.-
KELLI R. MANN, O . D .~
NORMAN W. BRAY, Ph .D.c
ABSTRACT: The near retinoscopy technique
of refractive error determination was compared
to the standard method of cycloplegic refraction
using 10 "infants" (3-1 2 months of age) and 10
"children" (32- 109 months of age). There was a
significant difference between the techniques for
both sphere and cylinder power. Although there
was no interaction of refractive technique and
age group, the difference between near retinos
copy and cycloplegic refractive error tended to
be larger for infanta than for children. No sig-
nificant difference was found when the average
refractive values were compared for monocular
or binocular conditions and no significant effect
was found for either gender or laterality (right
versus left eye). Based on these findings, it is
suggested that caution be used in substituting
the near retinoscopy technique for cycloplegic
refraction even utili:l.ing a "correction" factor fo r
the diopt ric difference between techniques.
KEY WORDS: near retinoscopy technique, cy
cloplegic refraction, infanta, children, refractive
error
Weuon MD, Mann KR, Bray NW. A comperilon or
cydoplegic refr.ction to the near retinollCOPY tech
nique for rerractiva e rror determination. J Am Optom
A.IIOC 1990; 61:680-4.
680 Journal or the American Optometric Association
technique for refractive error
determination
M Ohindra described a near retinoscopic technique as a substitute
for cycloplegic procedures determining refractive error.I - 2 This tech -
nique will be referred to herein as the "near retinoscopy technique."
The basic technique requires occlusion of one eye while retinoscopy
is performed on the open eye at a distance of 50 cm in a totally
dark room. In this manner the subject will presumably fixate on
the light of the retinoscope along the visualaxis.3 An adjustment in
lens power for both tonus of accommodation a nd working distance
from the retinoscope light is estimated to be 1.25 diopters. T his
value was empirically derived from a study co nducted on 27 adult
subjects 2 and again determined experime ntally by measuring accom-
modative responses in adults. The 1.25 diopters is subtracted from
the gross retinoscopy to yield t he final noncycloplegic value.
Since its inception, the near ret inoscopy technique has been
promoted for use with infants and young children to provide a
consistent refractive determination that compares favorably with
cycloplegic techniques. Mohindra originally reported that the tech-
nique had good reliability with infants using a comparison between
two examiners.' There were several aspects of Mohindra's study,
however, t hat were not clear. First, the age range of the subjects
was not. given. T his is important since variability in age may be a
factor in determining the efficacy of the procedure. Second, al-
though all subsequent studies of t his technique used monocular
occlusion, this one did not. However, it is possible that occlusion
was used but not described in the procedure. Finally, although
inspection of the published graphs appears to display good correia
tion between the two exam iners, no statistical analysis was per-
formed to verify this observation. The apparent variability of the
hyperopic values may have reduced any statistical correlation to a
much greater extent tha n can be deduced by a visual comparison of
the graphs.
To confirm t.he accuracy of this technique as a reliable substi-
tute for cycloplegic retinoscopy, Mohindra and Molinari compared
t he near ret.inoscopy method to cycloplegic refraction utilizing 3 1
subjects, 3- 7 years of age .~ Their results provided good interexam-
iner reliability fo r spherical and cylindrical power (r=+0.S3 and
+0.75, respectively) and good correlation between techniques for
spherical power (r=+0. 75) .
Unfortunately, the drugs used were 1% tropicamide (Mydriacyl)
and 10% phenylephrine hydrochloride (Neo-Synephrine). Tropi-
camide is not an appropriate agent for cycloplegic refraction since
it has been shown to leave an average 3.500. of residual accommo-
dation with as much as 6.250 . of accommodation in a sample
ra ngi ng from infancy to 9 years of age.s-1 Alt hough it has a rapid
onset, wit.h muimum cycloplegia occurring in 20-30 min, the effect
quickly diminishes.s Of course, phenylephrine hydrochloride has
no cycloplegic action. One other problem with Mohindra and Moli
nari's study was that it did not include infants, the literature suggests that this may be the case. Variability
group that this refractive technique was designed to may be less for the older population resulting in smaller
evaluate. differences between the near retinoscopy technique and
In a comparison study, Maino et aL evaluated a cycloplegic retinoscopy. The present study was designed
much Jarger (3 11 subjects) and younger sample (18-48 to investigate these possibilities.
monthsl.1I Their method of cycloplegia incorporated
0.75% tropicam ide combined with 2.5% phenylephrine. Method
The participants were first examined using the near
A total of 20 infants and young children from t he
retinoscopy technique of Mohindra and Molinari.s Re-
pediatric population at the Chauncey Sparks Center fo r
tinoscopy was performed 20 min after instillation of the
Developmental and Learning Disorders were used as
drops. Results indicated that t here was little agreement
subjects after obtaining the parents' informed consent.
between the two procedures in preschool children. Us-
The mean age for the overall group was 43.6 months
ing the 1.25 diopter correction factor, there was only a
(range: 3- 109 months; SD=42.4 months). There were
35.7 percent agreement between the cycloplegic values
nine males and 11 females. For purposes of further
and the adjusted near retinoscopy values. Other modi-
analysis the participants were grouped into two age
fications of the correction factor were attempted with
groups of 10 subjects: "i nfant s~ (range 3- 12 months;
only a marginal improvement in agreement. When they
mean==7 months, SO=3 months) and "child ren~ (range
evaluated children with larger refractive errors (>+3.00
32- 109 months; mean 80.2. SD=30.1 months).
O.S. or > -1.00 cylinder), there was even less agree-
The near retinoscopy technique was performed first
ment. There was agreement only 27 percent of the time
fo r all participants. Each participant was randomly
within + / -0.50 D.S. Although a large sample was used,
assigned to a counterbalanced order of binocular/ mo-
infants were not evaluated and the drugs used for
nocular or monocular/binocular procedures. Near reti-
cycloplegia suffered from the same problems as the
noscopy was rendered at a 50 cm test distance in a dark
Mohindra and Molinari study.
room by one examiner (KM) . Lens bars were used to
Borghi and Rouse addressed the issue of using a
find the neutral point which indicated a change from
better cycloplegic agent in t heir 1985 study. 1O They
"with" to "against." The fin al recorded value was ad-
compared the near retinoscopy met hod to cycloplegic
justed by subtracting the empirically derived value of
retinoscopy utilizing an older sample of 21 subjects
1.25 diopters used by Mohindra. u
whose ages ranged from 3.5- 10 years. One percent
Following the near retinoscopy technique, cyclople-
cyclopentolate was employed (2 drops in each eye re-
gia was achieved with 1 drop of 1% cyclopentolate
peated at 5 min intervals) . Cycloplegic refraction was
(Cyclogyl) in each eye fo llowed by a waiting period of
performed 35 to 40 min following the instillation of
45 min (+/ -5 min). Retinoscopy followed in the same
drops. On the average, 0.630 more plus power was
order as the near retinoscopy technique and was per-
measured by the cycloplegic method and these results
fo rmed at 66 cm. The final recorded value was adjusted
were significantly different for t he vertical, horizontal
by subtracting 1.50 diopters, the working distance cov-
and combined meridians (p=0.005; 0.0005; and 0.0001
ered to dioptric value.
respectively). The authors indicated that if one were to
"cut~ the cycloplegic findin g by +0.500. to + 1.000 .
Results
before prescribing, the results between the two proce-
duted would be in close agreement.
Refractive comparison
There are a nu mber of issues which remain to be
resolved. First, the technique would have its greatest Each eye of every subject in the two age groups was
utility with infants, but this age group has not been refracted a total of four times: Under monocular con-
studied directly. Although infants were evaluated fo r ditions with the near retinoscoy and cycloplegic tech-
interexaminer reliability in Mohindra's earliest study, niques, and under binocular conditions with the near
infants were not evaluated fo r cycloplegic/near retinos- retinoscopy and cycloplegic tech niques. Thus, on each
copy comparison in later studies. The Maino study age group, 80 refractive measurements were available
included children in the 18-48 month age range while for the right eye and for the left eye (10 subjects x 4
the Borghi and Rouse study included children 42- 120 refractions/eye/subject) for both sphere and cylinder
mont hs. measures. When considering the two refractive tech-
Second, it is possible that covering one eye is not a niques separately, a total of 40 refractive measurements
necessary component of t he procedure. The original were obtai ned for each technique.
paper did not use occlusion. I Further, Griffin aJludes to Sphe rical values were analyzed in a mixed analysis
t he possibility that occlusion does not make a differ- of variance (ANOVA) using t he P4V programs of
ence. 1I Finally, it is possible that the technique has BMDP. 12 The variables included age group (infants and
greater merit with older children than with infants. The children), gender, technique (near retinoscopy and cy-

Volume 61, Number 9, 9/90 681

cloplegic), ocular condition (monocula r and binocular),
eye (right and left), with repeated measures on the last
t hree factors. There was a significant effect fo r t ech-
nique, F(1 ,16)-13.58, p<.002, M8.-1.95, with means
of +0.53 (80:2.19) and +1.97 (80- 1.89) for t he near
retinoscopy a nd cycloplegic techniques, respectively.
This result indicates that when the data from all par-
ticipants are considered (infants and children com -
bined) , the near retinoscopy technique resulted in less
refractive error than the cycloplegic technique. There
was no s ignificant difference between age groups,
F(1, 16):1.22, p>.28, M5.:27.71 , with means of +1.62
(80:2.53), and +0.88 (80 =1.64) for the infants and
children, respectively.
As indicated by a nonsignificant interaction of tech-
nique a nd age group, F(1,16) = 2.93. p>. l O, MS. =1.95,
the same pattern holds fo r both infants and children.
The means were +0.56 (8 0 =2 .67) and +2.68 (SO: 1.89)
for near retinoscopy and cycloplegic for t he infant s and
+0.50 (80 :1.60) and +1.26 (SO=1.S2) for the near
retinoscopy and cycloplegic Cor t he children.
The main effect for ocular condition was not stat is-
t ically significant, F(l,16)=0.49, p>.49, M8. =.09, with
means of +1.23 (80=2. 18) and +1/ 27 (80= 2.15) for
monocular and binocular conditions, respectively.
There was no interactions with ocular condition a nd
age, F(1,IS)=.06, p>.81 , MS. =.09, with means of +1.59
(80 ""'2.57) and + 1.64 (50=2.52) for the monocular and
binocular condit ions for the infants and +0.87
(50=1.64) and +0.89 (80=1.66) Cor the monocular and
binocular conditions for the children. These results
indicate that t he effect oC covering one eye during the
objective refract ion does not influence the measures for
either age group.
One might speculate that there may be difference
if the data for the ocular conditions were treated sepa-
rately Cor the near retinoscopy and cycloplegic tech
niques. However, the interaction oC ocu lar condit ion
a nd technique was not statistically significant,
F(l,lS)=.14, p>.71, MS.=.004, with means Cor the mo
nocular and binocular conditions of >0.52 (50=2.22)
and +0.54 (50=2.18) with the near retinoscopy tech-
nique and +1.94 (50= 1.90) and +1.99 (50 =1.89) fo r
the cycloplegic technique. This is an indication that
t here is no efCect due to ocular condition for either the
near retinoscopy or cycloplegic technique.
The main effect for gender was not significant,
F(1 ,16)=.25, p> .S2, M5. =27.71, with means of +1.3 1
(50 =2.15) and +1.20 (50=2. 17) for the males and
females, respectively. T here was no interaction of gen-
der and age, F(l,IS)=1.21 , p>.28, M5.=27.71, with
means for the males and females of +2.60 (SO=2.20
and + 1.19 (80=2.5S), respectively, for the inCants and
+ 0.67 (50=1.83) a nd +1.20 (50=1.26), respectively,
for the children. T hese results indicate that there was
682 Journal of the American Optometric Association
no statistically significant effect for gender for either
age group.
T here was no difference between the right and left
eye as indicated by the nonsignifica nt main effect for
eye, F(1,lS)=.48, p>.49, M5.""'2.S1, with meansoC + 1.38
(80 =2.25) and + 1.12 (50 =2.07) for the right and left
eye, respectively. The interaction of eye and age was
not significant, F(l,16 )=.88, p>.3S, M8. =2.Sl. T he
means fo r the right and left eyes for the infants were
+ 1.16 (80=2.47) and + 1.63 (50= 2.62), respectively,
and for the children the corresponding means were
+1.14 (5 0 =2.00) and +0.S1 (80=1.14). This nonsig-
nificant main effect and interaction indicates that there
were no differences between eyes for either age group.
There were no other significant effects or higher order
interactions with the spherical measures.
Cylinde r powe r compa rison
Cylinder power was used as t he dependent variable in
a second mixed ANOVA with the same variables as
used in t he a nalysis of the spherical refractive error
data. As with spherical refractive error comparison,
there was an overall significant difference between the
two techn iques for cylinder power F(l,16)=10.25,
p<.OO5, M5.:: 1.25. The mean cylinder value for the
near retinoscopy method was -0.20 (50 =.36) while for
the cycloplegic it was -0.36 (50=.49). There was no
mean effect for age F(l,16):0.04, p>.85, MS. =1.25,
ocular condition (monocular and binocular),
F(l ,16)= 0.07, p>.79, MS e =.02, gender F(l ,16)=2.04,
p>. 17, M8.=1.25, and eye F(l,IS)=3.49, p>.08,
M5.= .15. All inte ractions were nonsignificant.
Discussion
The results indicate that 1) there is a significant differ
ence for both sphere and cylinder powe r between the
cycloplegic and near retinoscopy technique; 2) it makes
no difference whether the techniques a re performed
monocula rly or binocularly as Griffin has suggested,l l
3) there appears to be no significant difference as the
result of gender, age or eye tested.
Borghi and Rouse's results suggest little difference
(0.630 more plus in the cycloplegic than the near reti
noscopy method) between t echniques for their sample
although the differences were statistically significant.
This may be explained by the age of the subjects whose
ages ranged from 42- 120 months. The average age of
our total sample of 20 was 43.S months (80 =42.4) with
a maximum age of 109 months. Ten of those subject s
were 12 months or less with an average age of 7 months
(50=3), while the older children had an average age of
80.2 months (8D=30.1). Thus, any comparison between
our study a nd that of Borghi and Rouse should be
 Table I: Correlation coefficients: near 5.000 , - -- - -- - -- -
4.500
retinoscopy versus cycloplegic refraction 000
J.500
condition N correlation (R) P ....Iue J ooo
2.500
00 sphere 7.000
~ 1.500
monocular 20 .74 < .001 rtl 1,000
binocular 20 .75 <.001 ::. 0.500
("I 0,000
"
05 sphere tI 0.500
monocular 20 .63 <.002 1000
-1.500
binocular 20 .67 < .001 :rooo
00 cylinder
monocular
binocular
20
20
.90
.86
< .001
<.00 1
-:;1.500
O5.OOO'-::-::-:!::---:,-::--::-
-05.500 ~~ , ------t
-1.000 0 ,000 1000 2.000 05,000 4.000
,J
5.000
OS cylinder COMBSC
monocular 20 .80 <.001 FilfUre I : Spberical va lue. to r tbe le l't and rigbt eye unde r
monoc ul a r a nd b lnoc ul.r eon dition a were c omb ined
binocular 20 .03 <.001 (CQ MBS M:ane.r point retin oaoo py; COMBSCEClyc1opJegie r e_
tinOlCOpy). Their ....erage v.lues for each of tbe 20 aubjecta
ror th e two refrac ti ve metboda were tben co mpa red in th is
confined to the older group. For these children, we IlCaUerplo t. The co rrelation between me thod. wa. 0.68. The
found more plus in the cycloplegic refract.ion than ob- regreat;ion line Willi ploued witb a s lope of 0.1'\2 a nd the Y
served wit.h the near ret.inoscopy method (0.820). T his intercept , -1 .070.
is very similar to Borghi and Rouse's value of 0.630.
However, for our infant group, t.he difference between t ive correlations, this means that individuals who tend
the near retinoscopy and t.he cycloplegic technique was to display large values with cycloplegic agents also tend
2.120, nearly three times greater than the older popu- to display large values with the near retinoscopy tech-
lation, even t hough the interaction of age group and nique. Conversely, those with low cycloplegic values
techn ique failed to reach conventional levels of statis- will tend to be low with the near retinoscopy technique.
tical significance. Adding a correction constant such as +0.50 to +0.750
This suggests that using the near retinoscopy tech- to the near retinoscopy value will not change an indi-
nique for anything but a gross indication of refractive viduals ranking. After the correction factor, those with
error can lead to very tenuous conclusions concerning the greater values on one measure will still be relatively
refractive error. With this in mind, we reanalyzed this greater on the other measure while those low on one
data using all 20 subjects to see if the differences measure will also be low on the other. The average
between techniques reveal greater differences for my- scores fo r the group will be more similar with such a
opes or hyperopes. Difference calculations were made correction, but the correlation between measures will
only on t hose subjects displaying either myopia on all not change. This means t hat the error in prediction will
measures (n::8) or hyperopia on all measures (n=12). not change, only t he group average. These problems are
The average difference between techniques for myopia of even greater importance with the moderate correla-
was +2.61 (SO=1.79) while the difference for hyperopia tions which have a higher error of prediction. It is the
was +0.65 (50::.78). T his difference between myopia sphere measures that are utilized most, and fo r t hese
and hyperopia proved to be statistically significant there are only moderate correlations between t he near
F(1,16)=8.89, p<.009. However, due to the small sample retinoscopy and cycloplegic techniques. Figure 1 pro-
size under the constraint mentioned above, we were vides a scatterplot of combined measures comparing
unable to evaluate infants and ch ildren separately fo r near retinoscopy with cycloplegia for spherical values
these differences. under monocular and binocular conditions. A moderate
T he final point to be made from this data concerns correlation of 0.68 is obtained. T he square of t he cor-
the interpretation of the correlation between the near relation is.47 which means that the two methods have
retinoscopy values and those of cycloplegic refraction. only 47 percent of their variance in common, leaving
Table 1 displays the correlation between the techniques 53 percent unaccounted.
for sphere and cylinder values under monocular and One other complication is t hat we know that if the
binocular conditions. The correlations between the near retinoscopy results increase in magnitude over an
techniques range from moderate (.63) to high (.90) and age range, then the cycloplegic results also will increase.
all are sign ificantly greater than zero. Therefore, why However, the magnitude of the change depends on age
can the two techniques not be utilized interchangeably'? and t here is a t rend for t his difference to be greater fo r
The reason is that correlation only indicates the degree infants than for older children. Therefore, it is not
of consistency between individuals, not the magnitude appropriate to state that "on t he average" the cyclo-
of t he difference between measures. With st rong posi- plegic refraction will be +0.500 to +0.750 more plus

Volume 61 , Number 9, 9/90 683

than the near retinoscopy technique. Using the regres-
sion equations generated by these data will not be valid
because there is a significant difference in values be
tween techniques and there is a trend for the magnitude
of this difference to be larger for infants than for
younger children. It should be noted, however, that the
interaction of age group and technique failed to reach
conventional levels of statistical significance.
Conclusion
Contrary to other publis hed reports, our data indicates
that the near retinoscopy and cycloplegic techniques
yield significantly different values and can not be used.
interchangeably. There appears to be a trend towards
less difference in older children but the size of our
sample is too s mall to conclusively establis h th is trend.
We suggest that until much larger studies of infants
and young children a re carefu lly conducted, the refrac
tive results of the near retinoscopy technique should be
interpreted with caution.
References
I. Mohindrll I. A te<:hnique for infant e:lllminalion.t. Am J Optom
Ph)'lliol Opt 1975: 52:867-70.
2. Mohindra I. Compari!l(ln of Mn ear retinOKopyR . nd IUbjecti... e
refraction in adults. Am J Opt.om Phyaiol Ollt 1977; 54:3 19-22.
3. Mohindrl I. A non-cydoplegic refraction tl'( hnique for inflnu
.nd young children. J Am Optom Assoc 1977; 48:518-23.
4. Owens DA, Mohindra I, Held R. The effl'(li"'eness of I retino
ICope beam 118 I n .etommodllti ... e stimulus. Inve$t Ophthalmol
Vii Sci 1980; 19:942-9.
684 Journal of the Ame rican Optomet ric ASlIOCiation
5. Mohindra I, Moli nlri J f.'. Near retinoteopy and cyeloplellic reti-
nOllCOpy in early primary grade IChoo lchildren. Am J Optom
Phyaiol Opt 1979: 56:34-8.
6. Cett.es BC. T ropicamide, a new cycloplegic mydriat ic. Arch
Ophthalmol l96 1; 65:632- 5.
7. Milder B. T ropica mide 118 I cycloplegic agent. Arch Ophthalmol
1961: 66:70-2.
8. AmOfl OM. Cycloplegic refraction. In: Bartlett JD , Jaanu s SO
eca. Clinical ocular pharmacology. 2nd ed. Bosto n: Butterworth"
1989:42.5.
9. ,,"18ioo J H, Cibi, CW, Creu P, do .1. Noncycloplegic ..... cyclo
plegic retinOllCOpy in pre-Khool children. Ann Ophthalmol 1984;
16:880- 2.
10. Borghi RA , Rouse MW. ComperilOn of refraction obtained by
~ near retin OllCOpY and retinoteopy under cycloplegi . Am J
Opt.om Phy,io] Opt 1985: 62: 169-72.
It. Griffin JR. Binocular Anom.li": Pl"QCedure. for Vision Therapy.
2nd ed. Chic'go: proreuional Pre.., 1982:436.
12. Dizon W A (Ed.). BMDP Statistical Softw'n! . Berkely, CA: Uni
versity or C.lifornil Pre.., 1985.
Footnotes
a. A!I$OCi.te Proresso r ofOpto melry/Directo r, Di ... i, ion of Optometry,
Chauny Sperks Center for De ...elopmental .nd Leuning Di,
orden:, F.A.A.O.b. Optometrill./Residenl, Pedietric Optomet.ry.c.
Profeuor of J>.ychology/Director, Di... ision of P$yc hoIOl)',
Chauncey Sparks Ce n!.!! for De ... elopmental .nd Learn ing Di,-
orde rs
Submitted 11/89; accepted 5/90
University o( Alabama at Birmingham
Department of Optometry
School of Optometryl
Medical Center
Birmingham, AL 35294