Original Article

Random dot stereogram E in vision screening of children Carolyn Hope, FRACO Department of Ophthalmology, Auckland Public Hospital Keith Maslin, FRACO Department of Ophthalmology, Auckland Public Hospital

Abstract The random dot stereogram E (RDE) has been shown to be a simple and effective test for the detection of binocular abnormalities and defective visual acuity in ~hi1dren.l.~ We determined the validity of the RDE as a screening test for reduced visual acuity, amblyopia and strabismus in two separate populations of children.

A nonselective group of 100 school children (aged 5 to 15 years) who presented consecutively to the ophthalmology department at Auckland Public Hospital were tested with the RDE. All cases of amblyopia and strabismus were detected by the RDE. Similar screening with the RDE test of 168 preschool children (aged three to four years) in the community resulted in an unacceptably high over- referral rate. The test was unreliable in the preschool age group because of difficulty in distinguishing between test failure and non-cooperation with the test. The low positive predictive value of the test in the younger age group suggests the test to be unsuitable for preschool vision screening.

Key words: Random dot stereogram, vision screening.

Amblyopia remains an important cause of vision loss in the paediatric community. The reported prevalence varies owing to differences in populations studied and visual acuity criteria adopted. Amongst preschool and school-age children figures range between 1.3 and 3.5?k5

Early detection and correction of ocular defects may prevent permanent visual disability from amblyopia, and allow for the development of an optimal binocular function.6-8 The benefits of preschool screening are less well do~umented.~ At present detection of amblyopia is dependant largely upon the subjective measurement of visual acuity at three to four years when cooperation is possible. In New Zealand, preschool screening of amblyopia is primarily carried out by paediatric community health nurses (Plunket nurses), who test visual acuity using the four-metre letter matching test (LMT) without confusion bars,'O and binocular function using the Hirschberg corneal reflex test and cover test.

Single optotypes, easier to use in this age group, are inappropriate for amblyopia detection since children may pass at an artificially high level. This defect of isolated symbols may be overcome by surrounding the letter with adjacent black bars or other letters in order to activate the crowding effect.'O." The Hirschberg corneal reflex test and cover test are difficult to perform and may be unreliably interpreted in the young child. Visual acuity tests alone will fail to detect certain binocular anomalies such as alternating strabismus, high phorias, and hypermetropia. Instead, a stereotest, the random dot E (RDE) has been advocated as a screening test for identification of both reduced visual acuity and binocular anomalies.'-4.'2 Unlike other stereotests the RDE provides no monocular cues, and only when binocularly fused can the stereoscopic figure be seen.' Originally the RDE was used to detect amblyopia,' but recently it has

Reprint requests: Carolyn Hope, Department of Ophthalmology, Auckland Public Hospital, Grafton Road, Auckland, New Zealand.

Random dot stereogram E in vision screening of children 31 9

been proposed as an effective test in vision screening of ~ h i l d r e n . ~ ~ ~ ~ ’ ’ In one study of school-age children the RDE test was found to be more effective than other screening techniques.’

These encouraging reports prompted us to evaluate the RDE as a screening test for visual defects in two different populations of children - school children referred to the hospital ophthalmology department, and preschool children tested in the community.

Patients and methods Group 1 Group 1 consisted of 100 school children (aged five to 15 years) who presented consecutively with vision problems to the ophthalmology department at Auckland Public Hospital between March and June 1989. Each child was initially tested with the RDE at a distance of one and two metres by an independent examiner. The child then underwent a complete ophthalmic examination including Snellen acuity, binocular function, refraction and fundus examination. The RDE was carried out according to the test procedure and pass or fail was recorded for the one metre and two metre test distance. Refraction was carried out under cycloplegia in all children under seven years of age, and in older children where distance fixation was unreliable or where hypermetropia of greater than three dioptres was found with the initial retinoscopy. ‘For cycloplegia one drop of 1% cyclo- pentolate was used in each eye and this was repeated after five minutes. Retinoscopy was performed after a minimum period of 40 minutes. Refraction was recorded as cycloplegic refraction less 0.50 dioptres. Children who were found to have a visual acuity of 6/12 vision or less in either eye, or a manifest strabismus were recorded as positive referrals. The validity of the RDE as a screening test was then calculated for both the one and two metre test distances. We also evaluated the RDE in screening cases of amblyopia, manifest strabismus, anisometropia, and astigmatism. Amblyopia was defined in this study as a difference of two or more lines in Snellen acuity between eyes with correction, and prior to occlusion. Anisometropia was taken to be 1.50 dioptres or more difference between the corresponding axes of the two eyes.

Group 2 Group 2 consisted of 168 preschool children (aged three to four years) who attended for comprehensive

Table 1. Detection of ocular abnormalities by RDE at one metre in school children.

Reduced VA andlor manifest strabismus Results of RDE (1 m)

Present Absent Total

Positive (fail) 33 3 36 Negative (pass) 29 35 64 Total 62 38 100

Sensitivity of RDE at 1 m=33/62 (53%); Specificity of RDE at I m = 35/38 (92%); Positive predictive value = 33/36 (92%); Negative predictive value = 35/64 (55%).

health checks, including vision screening, conducted by Plunket nurses. In addition to vision screening tests routinely performed by the Plunket nurses, they also tested children with the RDE at a one and two metre test distance. We had origi- nally planned to examine all children who failed the RDE at the two metre test distance. However, since almost half the children failed at this test distance the criterion for pass or fail was made on the basis of the one metre test distance.

The following tests were administered to all preschool children. (Group 2)

1. Visual acuity - four-metre letter matching test (single optotype without confusion bars).’O Criteria for referral: 4/8 or less in either eye. or Allen picture cardst3 - six picture cards were presented to each eye at four metres. Criteria for referral: Those children who failed to correctly identify the pictures at four metres (equivalent to 6/15), or a shorter test distance with either eye.

2. Near cover test - All tropias were referred.

3. RDE - Criterion for referral: Failure to correctly identify the E at one metre at k s t four times in succession.

Table 2. Detection of ocular abnormalities by RDE at two metres in school children

Results of RDE (2 m) Reduced VA andlor manifest strabismus

Present Absent Total

Positive (failure) 53 18 71 9 20 29 Negative (pass)

Total 62 38 100

Sensitivity of RDE at 2 m=53/62 (85%); Specificity of RDE at 2 m = 20/38 (53%); Positive predictive value = 53/71 (75%); Negative predictive value = 20/29 (69%).

320 Australian and New Zealand Journal of Ophthalmology 1990; 18(3)

All children who failed one or more of the tests underwent a complete ophthalmic examination, which included a repeat visual acuity using the four- metre LMT without confusion bars, or Kaye picture cards,I4 if not cooperative with the letter matching test. Other tests included the RDE, binocular function, refraction, and fundus exami- nation. Those children with visual acuity of less than 4/4 with the LMT, or less than 6/6 with the Kaye picture cards in either eye underwent cycloplegic refraction as described for group 1 children. Children confirmed to have visual acuities of 6/12 or less in either eye or a manifest strabismus were considered positive referrals.

The sensitivity, specificity and predictive values of the RDE were then calculated in this group of children.

Description of Random dot stereogram E The RDE test comprises two random dot vecto- graphic stereograms, a pair of polarised glasses, and a demonstration model consisting of thick card printed to simulate a random dot pattern with a raised embossed E in the centre (see Figures 1, 2). One of the stereograms when viewed through the polarised glasses reveals a raised figure E similar to the demonstration model. The other appears blank with only a random series of dots visible. The RDE is calibrated such that when viewed at 28 cm the disparity is 900 arc seconds. By moving the test card farther away the disparity is reduced. At one metre the disparity is 252 arc seconds, and at 2 metres the disparity is 126 arc seconds. These thresholds are probably underestimates of actual threshold since moving the test target farther away reduces the visual angle of the test target, therefore the target is harder to see because of non- stereoscopic reasons as well as because of reduced disparity. Stereonorms for three to five year olds are estimated to be 126 arc Results of the test are recorded as pass or fail. A pass requires the E to be correctly identified at least four times in succession.

Results Results of RDE in Group 1, school children A total of 100 children were screened and the RDE test was evaluated at one and two metres.

Results at one metre. At one metre 36 children failed the RDE (test positive), and 64 passed (test negative). Of the children who failed, the test 33 had ocular abnormalities and of the children who

Random dot stereogram E in vision screening of children

Fig. 1.- The random dot stereogram E test consists of two random dot vectographic stereograms, one containing the stereo- graphic E, a demonstration model showing an embossed E, and

a pair of polarised glasses.

passed, 35 had a normal ophthalmic examination. However, 29 children passed despite reduced acuities (see Table 1).

The sensitivity of the RDE at one metre was calculated to be 5370, and the specificity to be 92%, with a positive predictive value of 9270, and a negative predictive value of 55%.

Results at two metres. At two metres, 71 failed (test positive), and 29 passed (test negative). Of the children who failed the test 53 were found to have ocular abnormalities (true positives), while 20 who passed had a normal ophthalmic examination (true negatives), (see Table 2).

The sensitivity of the RDE was calculated to be 8570, and the specificity to be 5370, with a positive predictive value of 75%, and a negative predictive value of 69%.

We reviewed those children who passed the RDE despite reduced visual acuity (false negatives). Of

Fig. 2.- Reverse sides of stereograms and demonstration model.

32 1

Table 3. Vision test results in selected Group 1, school children Uncorrected VA RDE (2 m) Refraction Corrected VA Number

R L 616 6112

6/12 615

615 619

616 619

R L pass Plano 616 616

fail + 1.50 616 615

pass Plano 615 615

fail Plano 616 619

+ 2.00D

Plano

- 0.501 - 2.00 X 30

- 1.001 - 2.00 X 30

the nine children who passed the RDE at two metres (see Table 2), eight were myopic (< 1.50 dioptres), and one child was hypermetropic. At one metre the test was less specific. Of the 29 children who passed (see Table l), 26 had myopia or myopic astigmatism with a maximum of 2.00 dioptres and 2.50 dioptres. In addition, three children were hypermetropic with a maximum error of 4.00 dioptres. The corrected acuities in all 38 children considered false negatives were 619 or better in each eye.

The RDE test detected all school-aged children with amblyopia (14 children) and manifest strabismus (nine children) at both 1 and 2 metres. There were 27 children with anisometropia. Of the 18 children with anisohypermetropia, one child passed at two metres (Table 3, no 1) and two children at one metre, (Table 3, no 1 and no 2). It was interesting to note that 14 of the remaining 15 children were amblyopic. Of the nine children with anisomyopia, one child passed at two metres (Table 3, no 3) and one passed at one metre (Table 3, no 4). The RDE detected all children with greater than 2.50 dioptres of astigmatism.

Results of R D E in Group 2, preschool children A total of 168 children were screened with the RDE (see Table 4). Eight children were uncooperative to all screening tests and have not been included

Table 4. Detection of vision defects in Group 2, preschool children

Result of RDE Vision defect

Present Absent Total

Positive 8 39 47 Negative 1 120 121 Total 9 159 168

Sensitivity = 8/9 (89%); Specificity = 1201159 (75%); Positive predictive value =8/47 (17%); Negative predictive value = 120/121 (99%).

in the study. Of the 168 children tested, 47 failed the RDE at one metre (test positive). Of these, seven children were found to have reduced visual acuities due to refractive errors, and one child had strabismus (see Table 5). Thirty-nine children had a normal ophthalmic examination. One child passed the RDE test despite defective acuity (see Table 5, no 8). Corrected acuity in the left eye using the L M T with and without confusion bars was 4/20 and 4/12 respectively. With correction, stereoacuity with the Frisby stereo platesI6 was 340 secs, and with the RDE 252 secs.

The sensitivity of the test was calculated to be 8870, specificity to be 75%, positive predictive value to be 1770, and negative predictive value to be 99%.

Discussion This study confirms the RDE to be a simple and quick test requiring no verbal response.

In school-aged children the test is well understood and reliably detects children with amblyopia, strabismus, and amblyopiogenic conditions such as significant hypermetropia, anisometropia and astig- matism. The RDE is highly sensitive at the two metre test distance but less specific with more children failing the test despite a normal ophthalmic examination. All children with emmetropia passed the test at both test distances, however a significant number of children also passed the test despite reduced acuities. The majority of these children had low myopia. This suggests screening with the RDE in isolation would be inadequate.

The RDE test was unreliable in preschool children and many children failed the test who were normal by other criteria. This high false-negative rate was due in part to a numer of children who failed to understand the instructions given with the test. Ingram also found the test was unreliable in testing preschool children and concluded a significant number of the children either would not attempt the test or obviously guessed.” In our study

322 Australian and New Zealand Journal of Ophthalmology 1990; 18(3)

Table 5. Examination of positive referrals from preschool screening Visual acuity Refraction

R L R L Management Case RDE (1 m)

418 418 414 4/20 4/8 414 619 414 6/18

418 418 414 4/20 4/12 418 6/24 4/12 6118

-0.501- 1.50 x 135 +2.50/+ 1 . 0 0 ~ 180

+ 0.50 D.S. - 5.001 - 1 .OO X 35 + 1.001 - 3.50 X 45

P11 - 0.50 X 90 PI/+ 1 . 0 0 ~ 180

+ 1.00 D.S. + 1.001+ l.0Ox 180

P11- 1.25X 10 + 2.251 + 0.75 x 180

+ 0.50 D.S. -5.001- 1 . 5 0 ~ 1 6 5 - 2.001 - 2.00 X 180

P11- 1 .OO X 90 + 2.001 - 3.00 X 180

+ 3.50 D.S. + 1.25/+ 1 . 7 5 ~ 180

Follow-up FOIIOW-UP Exotropia G 1 asses Glasses

Glasses Glasses

Follow-up

FOIIOW-UP

we also found it was difficult to make the distinction between a child failing the test, guessing or misun- derstanding the test requirements. Many parents felt their children performed poorly with the RDE because the test was given in distracting and noisy surroundings, however on retesting at the ophthalmology department with only the parent, child and tester present, the majority of children still failed this test despite a completely normal ophthalmic examination. We therefore believe the primary reason for our high overreferral rates was the specific three to four year old age group of children screened, and the recognised difficulty in obtaining reliable results in testing of this age group. In Reinecke and Simons original study, 87% of the 191 children were older than four years.' However, Simons, who compared different stereotests in 179 three to five year olds surprisingly showed only a 1% over-referral rate for RDE, with rescreening apparently eliminating all over-referrals. l 2

One preschool child in our study passed the RDE with ease at one metre despite apparent amblyopia. In Reinecke and Simons' original study,' no patient with more than two lines difference in visual acuity between eyes, or less than 6/12 in the worse eye with correction, passed the RDE at one metre.

There has been much debate over recent years concerning the justification for instituting widescale preschool vision screening programme^.'^,'^ Children with cosmetically obvious squints are usually detected at an earlier age. At presert screening for visual defects depends upon subjective tests at an age when cooperation is not always possible, and the search for a simple and effective test for vision screening continues. The RDE, proposed as a most promising test,20 in our study proved not only too unreliable in the preschool age group, but also failed to detect a child with amblyopia. It is likely that other screening tests will

Random dot stereogram E in vision screening of children

need to be devised which meet the requirements of early detection, minimal cooperation, speed and simplicity, yet high sensitivity and specificity. Photorefraction may be a realistic alternative.21.22

Conclusions The RDE in screening of school-aged children will reliably detect amblyopia and strabismus, but alone fails to identify other visual defects including myopia. In preschool children the test proves too unreliable and results in large numbers of false- positive referrals. The random dot stereogram E is unsuitable for screening of preschool children owing to the unreliable responses in this particular age group and not because of the inadequacy of the test itself in detecting significant visual problems.

Acknowledgements We would like to thank the Auckland Plunket Society and the Orthoptic department for their help with this study.

References 1. Reinecke RD, Simons K. A new stereoscopic test for

amblyopia screening. Am J Ophthalmol 1974;78:714-721. 2. Hammond RS, Schmidt P. A Random Dot E stereotest for

the vision screening of children. Arch Ophthalmol

3. Rosner J. The effectiveness of the Random Dot E as a preschool vision screening instrument. J Am Optom Assoc

4. Ruttum MS, Bence SM, Alcorn D. Stereopsis testing in a preschool vision screening program. J Pediatr Ophthalmol Strabismus 1986;23:298-302.

5. Von-Noorden GK. Burian von-Noorden's binocular vision and ocular motility. 3rd ed. St Louis: C V Mosby,

6. Jastrzebsk GB, Hoyt CS, Marg E. Stimulus deprivation amblyopia in children. Arch Ophthalmol

7. Atkinson J, Braddick 0, Wattam-Bell J, er a/. Photorefractive screening of infants and effects of refractive correction. Invest Ophthal Vis Sci (suppl) 1987;28(3):399.

1986; 104~54-60.

1978;40:1121-1124.

1985;211-212.

1984; 102: 1030-1034.

323

8. Parks M M . Single binocular vision. In: Duane TD, ed. Clinical Ophthalmology vol 1 , Philadelphia: J B Lippincott Co., 1988;12.

9. Feldman W, Milner RA, Sackett B, Gilbert S. Effects of preschool screening for vision and hearing on prevalence of vision and hearing problems 6-12 months later. Lancet 1980;2: 10 14- I0 16.

10. Parr JC. Clinical Assessment of visual acuity. Trans Ophthal

11. Atkinson J, Anker S, Evans C, McIntyre A. The Cambridge crowding cards for preschool visual acuity testing. In: Lenk- Schafer M, ed. Transactions of the Sixth International Congress, 1987:482-486.

12. Simons KA. Comparison of the Frisby, Random Dot E, T.N.O., and Randot Circles stereotests in screening and ofice use. Arch Ophthalmol 1981;99:446-452.

13. Allen HF. A new picture series for preschool vision testing. Am J Ophthalmol 1957;44:38-41.

14. Kaye H. New method ofassessing visual acuity with pictures. Br J Ophthalmol 1983;67:131-133.

SOC NZ 1981;33:157-167.

15. Simons K. Stereoacuity norms in young children. Arch Ophthalmol 198 1 ;99:439-445.

16. Hinchliffe HA. Clinical evaluation of stereopsis. Br J Orthoptics 1978;35:46-57.

17. Ingram RM, Holland ??, Walker C, er a/. Screening for visual defects in preschool children. Br J Ophthalmol

18. Stewart Brown SL, Haslum MN, Howlett B. Preschool vision screening: a service in need of rationalisation. Arch Dis Child 1988;63:356-359.

19. Ingram RM. Should preschool children be screened for visual defects? Trans Ophthalmol Soc UK 1985;104:646-647.

20. Ehrlich MI, Reinecke RD, Simons K. Preschool vision screening for amblyopia and strabismus: Programs, methods, guidelines. Surv Ophthalmol 1983;28: 145-163.

21. Atkinson J, Braddick OJ, Durden K, Watson PG, Atkinson S. Screening for refractive errors in 6-9 month infants by photorefraction. Br J Ophthalmol 1984; 68: 105-1 12.

22. Kaakinen KA, Kaseva HO, Teir H H . Two-flash photorefraction in screening of amblyogenic refractive errors. Ophthalmology 1987;94: 1036-1042.

1986;70:16-21.

324 Australian and New Zealand Journal of Ophthalmology 1990; 18(3)