Uncorrected Visual Acuity and Noncycloplegic Autorefraction Predict Significant Refractive Errors in Taiwanese Preschool Children

Uncorrected Visual Acuity and Noncycloplegic Autorefraction Predict Significant Refractive Errors in Taiwanese Preschool Children Yu-Hung Lai, MD,1,2,3,4 Han-Yi Tseng, MD,1 Hsin-Tien Hsu, PhD,5 Shun-Jen Chang, PhD,6 Hwei-Zu Wang, MD1,2 Purpose: To investigate the accuracy of uncorrected visual acuity (UCVA), stereopsis, and noncycloplegic autorefraction (NCAR) tests performed by vision-screening technicians and to determine the best referral criteria when using these methods to screen for significant refractive errors in preschool children. Design: Retrospective, case-control, and cross-sectional study. Participants: We reviewed 1000 records for a population-based preschool vision-screening program. The target conditions were defined as myopia ⱕ⫺3.0 diopters (D), hyperopia ⱖ4.5 D, astigmatism ⱖ2.0 D, and anisometropia ⱖ2.0 D. Methods: Receiver operating characteristic (ROC) curve was used to calculate optimal referral cutoff values. The examination results obtained by the vision-screening technicians were compared with those obtained by a pediatric ophthalmologist, which were considered the gold standard. Main Outcome Measures: The efficacies (sensitivity, specificity, positive predictive value, and negative predictive value) of different tests were evaluated. Results: In 7.0% (95% confidence interval [CI], 5.3– 8.7) of the children, at least 1 eye showed 1 of the target conditions. If only the right eyes were considered, the prevalence of target conditions was 4.2% (95% CI, 2.9 –5.5). The ROC curve analysis indicated that the NCAR cylinder test (cutoff value ⱖ0.875 D) was the best test for screening target conditions. With regard to age groups, UCVA ⱕ0.75 (Snellen equivalent) and ⱕ0.85 were the best referral criteria for ages ⱕ4 years and ⱖ5 years, respectively. Combining the UCVA test with the NCAR test (the child was referred after failing both tests) increased specificity without significantly decreasing sensitivity. Conclusions: The UCVA and NCAR tests performed by vision-screening technicians are adequately sensitive and specific for preschool vision screening. The ROC curve analysis was used for determining the appropriate screening criteria for these tests, and combining the tests increased their accuracy. The screening criteria should be age dependent. When analyzing the test accuracy in ophthalmic problems, if the disease of interest does not symmetrically (in terms of disease severity and prevalence) involve both eyes, the prevalence based on only 1 eye should be interpreted with caution. Financial Disclosure(s): The author(s) have no proprietary or commercial interest in any materials discussed in this article. Ophthalmology 2013;120:271–276 © 2013 by the American Academy of Ophthalmology.

Many healthcare organizations recommend vision screening for preschool children to detect and treat pediatric eye and vision problems.1– 4 The rationale for vision screening in this age group is to detect vision-threatening problems, such as amblyopia, strabismus, or significant refractive error at an early age when visual plasticity is still high. In addition, a significant portion of preschool children have visual impairment, such as uncorrected refractive errors and amblyopia, which require ophthalmic care.5 Several preschool vision-screening projects have been proposed.1,4,6,7 Although many studies have reported “predictive values” for the accuracy of procedures used in preschool vision screening, the subjects are typically children who are referred after failing the screening and do not © 2013 by the American Academy of Ophthalmology Published by Elsevier Inc.

include the children who pass the screening. A thorough analysis of any screening program ideally requires comparison of its pass/fail rate with a gold standard,4 such as that used in studies of photoscreening.8,9 Moreover, there are few studies that are population-based.10 Although no screening method is 100% accurate, by selecting the proper tests, we can correctly identify abnormal subjects (i.e., minimize false negatives) and avoid the incorrect identification of normal subjects (i.e., minimize false positives) (Table 1). Recommended screening procedures include monocular distance visual acuity, stereopsis, the cover-uncover test, auto-refractor/retinoscopy, and photoscreening. Combining different screening tests to increase screening yield has been ISSN 0161-6420/13/$–see front matter http://dx.doi.org/10.1016/j.ophtha.2012.08.009

271

Ophthalmology Volume 120, Number 2, February 2013 Table 1. Definitions of Sensitivity, Specificity, Positive Predictive Value, and Negative Predictive Value

Visual Acuity Testing

Disease Status Test Results

Abnormal

Normal

Positive (fail) Negative (pass)

A C

B D

A ⫽ true positives; B ⫽ false positives; C ⫽ false negatives; D ⫽ true negatives; sensitivity ⫽ A/(A⫹C); specificity ⫽ D/(B⫹D); positive predictive value (PPV) ⫽ A/(A⫹B); negative predictive value (NPV) ⫽ D/(C⫹D).

suggested for patients with cancer.11,12 In developed countries, photoscreening is reportedly effective for vision screening.13 In most areas of Taiwan, however, photoscreening devices and trained operators and technicians may not be readily available. In such cases, monocular distance visual acuity, stereopsis, or the auto-refractor test is used. Therefore, for preschool children with high refractive errors, this study aimed to (1) test the accuracy of uncorrected visual acuity (UCVA), stereopsis, and noncycloplegic autorefraction (NCAR) tests performed by vision-screening technicians in a vision-screening program in Taiwan; (2) determine whether the accuracy of the tests varies with age; (3) determine the best cutoff values for these screening methods; and (4) test the applicability of different combinations of these screening tests in a preschool vision-screening program in Taiwan.

Materials and Methods Study Design All records were reviewed for the vision-screening tests performed for Taiwanese preschool children during 2005–2008. Some of the reviewed data, including the testability of the screening tests, has been published.7,14,15 All preschools actively volunteered to be involved in the preschool vision-screening program of Kaohsiung Medical University Hospital. The population-based screening program was designed to evaluate the accuracy of current methods of screening (visual acuity, stereopsis, and autorefraction) as performed by vision-screening technicians. This review of medical records adhered to the Declaration of Helsinki and was approved by the institutional review board of Kaohsiung Medical University Hospital, Kaohsiung, Taiwan. The subjects were aged 2 to 7 years and enrolled in preschools in Kaohsiung, Taiwan.

Screening Personnel The vision-screening tests included, in sequential order, NCAR, stereopsis, and UCVA. The tests were performed by optical technicians and vision-screening technicians from the Department of Ophthalmology, Kaohsiung Medical University Hospital. Their primary responsibility is to perform autorefraction tests and measure the visual acuity of patients in the hospital. Although a national board certification system has not been established for optical technicians or optometrists in Taiwan, the 2 participating technicians had 7 and 10 years (as of 2005) full-time work experience at this hospital with appropriate continuing education.

272

The Tumbling E chart distance visual acuity test (5 optotypes in a line; testing distance: 6 m; Taiwan Instrument Co., Taipei, Taiwan) was performed as previously described.7 Briefly, a binocular pretest was done using an optotype size of 0.1. For children undergoing the test, the left eye was covered to test the right eye beginning at line 0.2 (20/100). The UCVA was recorded as the smallest size at which at least 3 optotypes were correctly identified.

Stereopsis Test Stereopsis was evaluated using both the Titmus test and the National Taiwan University (NTU) stereopsis test, which is a random dot test that detects stereopsis of 300 seconds of arc or better.16 The test requires the subject to identify various shapes (i.e., circles, triangles, squares, and diamonds) in a random-dot image while looking through blue and red filters with the right and left eyes, respectively. A subject passes the test by correctly identifying 5 consecutive presentations. The results were recorded as “pass” or “fail.”

Refraction The vision-screening technicians used a table-mounted KR-8100 autorefractor (Topcon, Tokyo, Japan) to detect NCAR in each subject. The average value obtained over 3 tests (automatically calculated by the autorefractor) was used for analysis. Astigmatism was recorded in plus form.

Gold Standard For each subject, an eye-alignment examination, a Brückner test (corneal light reflex was tested to detect strabismus in younger and uncooperative children, and red reflex was observed to detect ocular media opacity, significant anisometropia, and possible small-angle strabismus), best-corrected visual acuity, cycloplegic retinoscopic refraction,7 or direct ophthalmoscopy (for those with unexplained poor best-corrected visual acuity) was performed by a single pediatric ophthalmologist (Y-H.L) on the same day. The results obtained by the ophthalmologist were used as the gold standard for examining and analyzing the screening results obtained by the vision-screening technicians.

Target Conditions The target conditions were significant refractive errors in either eye, including myopia ⱕ⫺3.0 diopters (D), hyperopia ⱖ4.5 D, astigmatism ⱖ2.0 D, or anisometropia with a spherical equivalent (SE) difference of ⱖ2.0 D or a difference in astigmatism of ⱖ2.0 D.3,17 The pediatric ophthalmologist (Y.H.L.) evaluated all subjects for the presence of the target conditions.

Receiver Operating Characteristic Curve Analysis For each target condition, receiver operating characteristic (ROC) curves were preliminarily used to determine optimal referral cutoff values for each test (UCVA, stereopsis, and NCAR) performed by the vision-screening technicians.18 Visual acuity was transformed into a logarithm of the minimum angle of resolution format for further analysis. An earlier study by the authors demonstrated that children aged ⱖ5 years have better visual acuity, as measured by the E chart, than those aged ⬍5 years.15 Therefore, ROC curves for the UCVA (including interocular difference) and stereopsis vision-screening tests were constructed for 2 different age groups (⬍5-year-old group and ⱖ5-year-old group). Because astigmatism reportedly has no significant association with cycloplegia19 –23 and the results of SE obtained by autorefractor will vary unless cycloplegia is used,24 only the cylinder values for NCAR (NCAR

Lai et al 䡠 Preschool Vision Screening by Autorefractor cylinder test) were used for the ROC curve analysis. The ROC curve for UCVA and NCAR were first examined in each eye to ascertain whether it differed between the right and left eyes. Tests with the best values of the area under the curve (AUC) were selected for further referral criteria analyses.

(i.e., ⱖ0.875 D). Therefore, for UCVA and NCAR, it was reasonable to use the same cutoff for both eyes. The test validities of the UCVA test (for ages ⱕ4 years, ⱕ0.75; for ages ⱖ5 years, ⱕ0.85), the NCAR cylinder test (cylinder power ⱖ0.875 D), and the NTU stereopsis test were examined.

Establishing Referral Criteria

Accuracy Test of New Referral Criteria

For each test, the best cutoff value for identifying high refractive errors was determined. On the basis of the ROC curve analysis, the selected referral criteria for UCVA, NCAR, and stereopsis tests were used to analyze the accuracy (sensitivity, specificity, positive predictive value [PPV], and negative predictive value [NPV]) of the criteria used to screen children with a target condition in at least 1 eye. The sensitivity, specificity, PPV, and NPV of the screening tests for the target conditions were calculated according to the formulas shown in Table 1. The tests also were performed in different combinations to investigate whether the combinations improved specificity without decreasing sensitivity or vice versa. The McNemar chi-square test was used for pairwise comparison of sensitivity and specificity between screening tests (SPSS 14.0, SPSS Inc, Chicago, IL).9 P⬍0.05 was considered statistically significant. The Bonferroni correction was used for multiple comparisons.

For children aged ⱕ4 years using a single test, the UCVA test (any eye ⱕ0.75) and the NCAR cylinder test (any eye cylinder ⱖ0.875 D) had the highest sensitivities (0.81 and 0.85, respectively), whereas the NTU stereopsis test, UCVA (any eye ⱕ0.75), and NCAR cylinder tests (any eye ⱖ0.875 D) had the highest specificities (0.82, 0.78, and 0.85, respectively). When the UCVA test (any eye ⱕ0.75) was combined with the NCAR cylinder test (any eye ⱖ0.875 D; the child was referred after failing both tests), the specificity (0.92) increased without significantly decreasing sensitivity (0.79). In single-instrument tests of children aged ⱖ5 years, the UCVA test (any eye ⱕ0.85) and the NCAR cylinder test (any eye ⱖ0.875 D) had the highest sensitivities (0.92 and 0.87, respectively) and the NTU stereopsis test had the highest specificity (0.93). When the UCVA test (any eye ⱕ0.85) was combined with the NCAR cylinder test (any eye ⱖ0.875 D; the child was referred after failing both tests), the specificity increased (0.91) without significantly decreasing sensitivity (0.89). Table 3, Table 4 (available at http://aaojournal.org), and Table 5 (available at http://aaojournal.org) show the sensitivities, specificities, PPVs, and NPVs for the screening tests evaluated in this study.

Results Demographic Data For 10 children, the data for age and gender were missing; therefore, these were excluded from the analysis. The average age of the remaining 1000 children (506 male and 494 female) enrolled in the study was 5.0 years (standard deviation [SD], 0.8 years; range, 2.5–7.3 years; 95% confidence interval [CI], 4.9 –5.0). The mean SE was ⫹0.43 D (SD, 1.04 D; range, ⫺7.38 to 6.88 D; 95% CI, 0.36 – 0.49) in the right eye and ⫹0.40 D (SD, 1.13 D; range, ⫺8.00 to 8.13 D; 95% CI, 0.32– 0.47) in the left eye. Mean astigmatism was 0.50 D (SD, 0.53 D; range: 0 –3.75 D; 95% CI, 0.46 – 0.53 D) in the right eye and 0.49 D (SD, 0.51 D; range: 0 –3.75 D; 95% CI, 0.45– 0.52 D) in the left eye.

Prevalence of Target Conditions In 7.0% of the children (SD, 0.3%; 95% CI, 5.4 – 8.7), at least 1 eye showed significant refractive errors, that is, 1 or more of the target conditions. If only the right eyes were considered, the prevalence of target conditions was 4.2% (SD, 0.2%; 95% CI, 2.9 –5.5); if only the left eyes were considered, the prevalence of target conditions was 4.4% (SD, 0.2; 95% CI, 3.0 –5.7). Anisometropia (a difference of ⱖ2 D) was found in 2.0% (SD, 0.1%; 95% CI, 1.1–2.9) of the children. In 0.6% of the children (SD, 0.1%; 95% CI, 0.1–1.0), at least 1 eye had myopia ⱕ⫺3.0 D. One or both eyes had hyperopia of ⱖ4.5 D in 1.0% of the children (SD, 0.1%; 95% CI, 0.4 –1.7), and 1 or both eyes had astigmatism ⱖ2.0 D in 5.1% of the children (SD, 0.2%; 95% CI, 3.7– 6.6).

Receiver Operating Characteristic Curve Analysis Table 2 presents the AUC, the P values for the ROC curves, and the best cutoff values for the tests. The UCVA test produced the most desirable AUC, and the best cutoff values for the left and right eyes were similar: ⱖ0.13 for ages ⱕ4 years (ⱕ0.75 [20/26.7], Snellen equivalent) and ⱖ0.07 for ages ⱖ5 years (ⱕ0.85 [20/23.5], Snellen equivalent). For all ages, the NCAR cylinder test produced the most desirable AUC and the same optimal cutoffs for the left and right eyes

Discussion The present results demonstrated that the NCAR cylinder test was the best method when using a single-instrument test to identify target conditions, providing the largest AUC and highest sensitivity/specificity, especially for children aged 4 years or less. One possibility is that the autorefraction produces more reliable data than the visual acuity test in such young children. The accuracy of the Topcon KR-8100 autorefractor, which is widely used in Asia for vision screening, has not been reported. Other autorefractor models are reportedly effective for detecting significant refractive errors.9,25,26 For example, at a specificity of 0.90, the Power Refractor II (Plusoptix, Nuremberg, Germany), Retinomax autorefractor (Nikon, Inc., Melville, NY), and SureSight (Welch Allyn, Inc., Skaneateles Falls, NY) autorefractor have minimum sensitivities of 0.43, 0.55, and 0.68, respectively.9,27 The sensitivity and specificity obtained with the Topcon KR-8100 autorefractor were comparable to those for the above models. Thus, sensitivity and specificity do not seem to differ substantially among these autorefractor models. This study also showed that the referral criteria need to be adjusted for age when using UCVA for preschool vision screening. Uncorrected visual acuity is probably the best single-instrument test in developing countries or areas with low resources, especially for children aged 5 years or more. The specificity increases and sensitivity decreases as the criterion used as the referral cutoff value decreases.9,28 A strict criterion such as UCVA ⱕ0.85 produces good sensitivity, as observed in the current study. When the specificity approached 0.9 (data not shown), the results were similar to those obtained in the Vision In Preschoolers study9 and another study in Eastern Taiwan (a mainly indigenous pop-

273

Ophthalmology Volume 120, Number 2, February 2013 Table 2. Results for Receiver Operating Characteristic Curve Analysis of Referral Criteria for Target Conditions*

Age ⱕ4 yrs RE UCVA for RE with at least 1 target condition LE UCVA for LE with at least 1 target condition IOD in UCVA for at least 1 eye with a target condition Titmus test for at least 1 eye with a target condition Age ⱖ5 yrs RE UCVA for RE with at least 1 target condition LE UCVA for LE with at least 1 target condition IOD in UCVA for at least 1 eye with a target condition Titmus test for at least 1 eye with a target condition All ages RE cylinder power for RE with at least 1 target condition LE cylinder power for LE with at least 1 target condition Titmus test for at least 1 eye with a target condition

AUC

P

95% CI

Best Cutoff

0.899

⬍0.001

0.842–0.956

ⱖ0.13 (ⱕ0.75 [20/26.7], Snellen equivalent, failed 20/25)

0.874

⬍0.001

0.784–0.965

ⱖ0.13 (ⱕ0.75 [20/26.7], Snellen equivalent, failed 20/25)

0.763

⬍0.001

0.663–0.863

ⱖ0.05 (⬃0.5 line, Snellen equivalent)

0.669

0.001

0.577–0.761

ⱖ90 seconds of arc (Circle Test #6)

0.941

⬍0.001

0.912–0.971

ⱖ0.07 (ⱕ0.85 [20/23.5], Snellen equivalent, failed 20/22)

0.888

⬍0.001

0.798–0.978

ⱖ0.07 (ⱕ0.85 [20/23.5], Snellen equivalent, failed 20/22)

0.657

0.008

0.530–0.783

ⱖ0.05 (⬃0.5 line, Snellen equivalent)

0.644

0.008

0.553–0.736

ⱖ55 seconds of arc (Circle Test #8)

0.945

⬍0.001

0.900–0.990

ⱖ0.875 D

0.912

⬍0.001

0.844–0.981

ⱖ0.875 D

0.668

⬍0.001

0.603–0.732

ⱖ70 seconds of arc (Circle Test #7)

AUC ⫽ area under the curve; CI ⫽ confidence interval; D ⫽ diopter; IOD ⫽ interocular difference in UCVA; LE ⫽ left eye; logMAR ⫽ logarithm of the minimum angle of resolution; RE ⫽ right eye; UCVA ⫽ uncorrected visual acuity (in logMAR). Circle Test #6 ⫽ Titmus test 80 seconds of arc; Circle Test #7 ⫽ Titmus test 60 seconds of arc; Circle Test #8 ⫽ Titmus test 50 seconds of arc. *Target conditions: hyperopia ⱖ4.5 D, myopia ⱕ⫺3.0 D, astigmatism ⱖ2.0 D, and anisometropia ⱖ2.0 D.

ulation).16 A recent study also successfully used ROC curves to optimize the visual acuity threshold for myopia screening in adolescents.29 Therefore, the optimal cutoff value cannot be generalized and depends on the purpose and the target population of the test. Stereopsis tests reportedly have an unacceptably low sensitivity when used as a single-measure test for detecting amblyopia, strabismus, or a high risk of refractive error.9,16,30 In addition, combining the stereopsis test with other screening devices and methods does not significantly improve its sensitivity.9 The results of the present study were consistent with those of the previous studies in this respect. Uncorrected visual acuity resulted in an excessive number of false positives when using a high cutoff value as the only referral criterion. Combining tests has improved screening yields in patients with cancer.11,12 In the current study, the UCVA test combined with the NCAR cylinder test (the referral criterion being failure in both tests) improved the specificity without significantly decreasing sensitivity. High cylinder values are reportedly associated with high ametropia,31,32 whereas both high cylinder values and ametropia are associated with poor visual acuity. This study recommended combining UCVA with the NCAR cylinder test to reduce unnecessary referrals in preschool children. Many years of training and experience are required to master the skills needed to perform the best-corrected visual acuity test, the cover-uncover test, and the cycloplegic (retinoscopic) refraction test. A reliable system also is needed to monitor and evaluate the performance of these tests by

274

vision-screening personnel. In addition, Taiwan has not established a national board certification system for optometrists or orthoptists. Compared with the best-corrected visual acuity test, cover-uncover test, and cycloplegic (retinoscopic) refraction test, the UCVA and NCAR tests are less time-consuming and easier to master for vision-screening technicians. After completing appropriate continuing education programs, the vision-screening technicians at Kaohsiung Medical University obtained sensitivities comparable to those reported in the Vision In Preschoolers study.9 This study suggests that vision-screening technicians can play an important role in preschool vision-screening programs in developing countries or cities with low resources. Another important finding in this study was that the prevalence of target conditions was different when it was reported for an eye (right eye, 4.2%; left eye, 4.4%) or for an individual (at least 1 eye with 1 of the target conditions, 7.0%). The proportional difference between the prevalence was approximately 60%. It has been well discussed that the characteristics of the right eye are more similar to those of the left eye of the same person than those of the eyes of another individual.33 The use of data from only 1 eye per person is statistically valid when a disease symmetrically involves both eyes.33 However, a significant portion of children had anisometropia (⬃2%; in this study, we used stricter criteria of a difference of 2 D); therefore, analyzing the prevalence in only 1 eye can underestimate the impact of the disease in a population. In addition, the prevalence affects the predictive values of tests. For instance, in a

Lai et al 䡠 Preschool Vision Screening by Autorefractor Table 3. Sensitivity, Specificity, Positive Predictive Value, and Negative Predictive Value for Screening Tests for Targeted Conditions Referral Criteria Age ⱕ4 yrs (n ⫽ 495) 1. NTU random dot test 2. Titmus Circle Test (#6) 3. UCVA ⱕ0.75 4. NCAR cylinder test ⱖ0.875 D 5. NTU random dot test or UCVA ⱕ0.75* 6. NTU random dot test and UCVA ⱕ0.75† 7. NTU random dot test or NCAR cylinder ⱖ0.875 D* 8. NTU random dot test and NCAR cylinder ⱖ0.875 D† 9. UCVA ⱕ0.75 or NCAR cylinder ⱖ0.875 D* 10. UCVA ⱕ0.75 and NCAR cylinder ⱖ0.875 D† Age ⱖ5 yrs (n ⫽ 505) 1. NTU random dot test 2. Titmus Circle Test (#8) 3. UCVA ⱕ0.85 4. NCAR cylinder ⱖ0.875 D 5. NTU random dot test or UCVA ⱕ0.85* 6. NTU random dot test and UCVA ⱕ0.85† 7. NTU random dot test or NCAR cylinder ⱖ0.875 D* 8. NTU random dot test and NCAR cylinder ⱖ0.875 D† 9. UCVA ⱕ0.85 or NCAR cylinder ⱖ0.875 D* 10. UCVA ⱕ0.85 and NCAR cylinder ⱖ0.875 D†

Sensitivity

Specificity

PPV

NPV

0.47 0.62 0.81 0.85 0.81 0.44 0.92 0.42 0.92 0.79

0.82 0.68 0.78 0.85 0.66 0.94 0.78 0.98 0.74 0.92

0.17 0.14 0.23 0.33 0.16 0.38 0.27 0.69 0.23 0.44

0.95 0.96 0.98 0.98 0.98 0.95 0.99 0.95 0.99 0.98

0.20 0.73 0.92 0.87 0.92 0.16 0.91 0.09 0.94 0.89

0.93 0.49 0.76 0.85 0.73 0.96 0.83 0.99 0.72 0.91

0.15 0.09 0.18 0.30 0.16 0.19 0.28 0.50 0.16 0.36

0.94 0.96 0.99 0.99 0.99 0.95 0.99 0.94 0.99 0.99

D ⫽ diopter; NCAR ⫽ noncycloplegic autorefraction; NPV ⫽ negative predictive value; NTU ⫽ National Taiwan University; PPV ⫽ positive predictive value; UCVA ⫽ uncorrected visual acuity (Snellen equivalent). Titmus Circle Test (#6) ⫽ ⱖ90 seconds of arc; Titmus Circle Test (#8) ⫽ ⱖ55 seconds of arc. *Combination of 2 tests: The patient should be referred after failing either test. † Combination of 2 tests: The patient should be referred after failing both tests.

population of 1000 persons, a test with a sensitivity of 90% and specificity of 90%, the PPV changes from 8.3% to 50% when the prevalence increases from 1% to 10%. Therefore, if a disease does not symmetrically (both severity and prevalence) involve both eyes, we should not report the prevalence and analyze the test accuracy from only 1 eye. In conclusion, the strengths of the current study are its large population size and thorough accuracy testing that was performed for all the screened children. We have provided a strategy for establishing preschool vision screening and verifying the screening criteria by using various test combinations and the ROC curve, which would help institutes with lower resources and equipment. A limitation of the study is the inclusion of only kindergarten and preschool children, who tend to have good visual acuity.34 A national survey in 2000 reported that approximately 96% of 5-yearolds in Taiwan were enrolled in preschools/kindergartens. In addition, the involvement of the preschools in this visionscreening program was not randomized and was limited to Kaohsiung. However, there was no evidence that the children in those preschools were significantly different from other parts of Taiwan, except in the Eastern part of Taiwan,16 where most of the inhabitants are indigenous.

2.

3.

4.

5. 6. 7. 8.

References 1. Hartmann EE, Dobson V, Hainline L, et al, Maternal and Child Health Bureau, National Eye Institute Task Force on Vision

9.

Screening in the Preschool Child. Preschool vision screening: summary of a task force report. Pediatrics 2000;106:1105–16. Committee on Practice and Ambulatory Medicine of American Academy of Pediatrics, Section on Ophthalmology of American Academy of Pediatrics, American Association of Certified Orthoptists, American Association for Pediatric Ophthalmology and Strabismus, American Academy of Ophthalmology. Eye examination in infants, children, and young adults by pediatricians. Pediatrics 2003;111:902–7. American Academy of Ophthalmology Pediatric Ophthalmology/Strabismus Panel. Preferred Practice Pattern Guidelines. Amblyopia. San Francisco, CA: American Academy of Ophthalmology; 2007. Available at: http://one.aao.org/CE/ PracticeGuidelines/PPP.aspx. Accessed July 31, 2012. Hartmann EE, Bradford GE, Chaplin PK, et al, PUPVS Panel for the American Academy of Pediatrics. Project Universal Preschool Vision Screening: a demonstration project. Pediatrics 2006;117:e226 –37. Pai AS, Wang JJ, Samarawickrama C, et al. Prevalence and risk factors for visual impairment in preschool children the Sydney Paediatric Eye Disease Study. Ophthalmology 2011;118:1495–500. Simons K. Preschool vision screening: rationale, methodology and outcome. Surv Ophthalmol 1996;41:3–30. Lai YH, Hsu HT, Wang HZ, et al. The visual status of children ages 3 to 6 years in the vision screening program in Taiwan. J AAPOS 2009;13:58 – 62. Tong PY, Bassin RE, Enke-Miyazaki E, et al. Screening for amblyopia in preverbal children with photoscreening photographs: II. Sensitivity and specificity of the MTI photoscreener. Ophthalmology 2000;107:1623–9. Schmidt P, Maguire M, Dobson V, et al, Vision in Preschoolers Study Group. Comparison of preschool vision screening tests as

275

Ophthalmology Volume 120, Number 2, February 2013

10. 11.

12.

13.

14. 15. 16. 17.

18.

19.

20. 21.

administered by licensed eye care professionals in the Vision In Preschoolers Study. Ophthalmology 2004;111:637–50. Chou R, Dana T, Bougatsos C. Screening for visual impairment in children ages 1-5 years: update for the USPSTF. Pediatrics 2011;127:e442–79. Collins JF, Lieberman DA, Durbin TE, Weiss DG, Veterans Affairs Cooperative Study #380 Group. Accuracy of screening for fecal occult blood on a single stool sample obtained by digital rectal examination: a comparison with recommended sampling practice. Ann Intern Med 2005;142:81–5. Berg WA, Blume JD, Cormack JB, et al, ACRIN 6666 Investigators. Combined screening with ultrasound and mammography vs mammography alone in women at elevated risk of breast cancer. JAMA 2008;299:2151– 63. Longmuir SQ, Pfeifer W, Leon A, et al. Nine-year results of a volunteer lay network photoscreening program of 147 809 children using a photoscreener in Iowa. Ophthalmology 2010; 117:1869 –75. Lai YH, Hsu HT, Wang HZ, et al. Astigmatism in preschool children in Taiwan. J AAPOS 2010;14:150 – 4. Lai YH, Wang HZ, Hsu HT. Development of visual acuity in preschool children as measured with Landolt C and Tumbling E charts. J AAPOS 2011;15:251–5. Chang CH, Tsai RK, Sheu MM. Screening amblyopia of preschool children with uncorrected vision and stereopsis tests in Eastern Taiwan. Eye (Lond) 2007;21:1482– 8. Donahue SP, Arnold RW, Ruben JB, AAPOS Vision Screening Committee. Preschool vision screening: what should we be detecting and how should we report it? Uniform guidelines for reporting results of preschool vision screening studies. J AAPOS 2003;7:314 – 6. Peat JK, Barton B. Medical Statistics: A Guide to Data Analysis and Critical Appraisal. In: Peat J, Barton B, eds. Categorical and continuous variables: diagnostic statistics. 1st ed. Malden, MA: Blackwell Publications; 2005:286 –7. Twelker JD, Mutti DO. Retinoscopy in infants using a near noncycloplegic technique, cycloplegia with tropicamide 1%, and cycloplegia with cyclopentolate 1%. Optom Vis Sci 2001; 78:215–22. Jorge J, Queiros A, Gonzalez-Meijome J, et al. The influence of cycloplegia in objective refraction. Ophthalmic Physiol Opt 2005;25:340 –5. Lin LL, Shih YF, Hsiao CH, et al. The cycloplegic effects of cyclopentolate and tropicamide on myopic children. J Ocul Pharmacol Ther 1998;14:331–5.

22. Liang CL, Hung KS, Park N, et al. Comparison of measurements of refractive errors between the hand-held Retinomax and on-table autorefractors in cyclopleged and noncyclopleged children. Am J Ophthalmol 2003;136:1120 – 8. 23. Liang CL, Hung KS, Park N, et al. Comparison of the handheld Retinomax K-Plus2 and on-table autokeratometers in children with and without cycloplegia. J Cataract Refract Surg 2004;30:669 –74. 24. Cowen L, Bobier WR. The pattern of astigmatism in a Canadian preschool population. Invest Ophthalmol Vis Sci 2003; 44:4593– 600. 25. Cordonnier M, Dramaix M. Screening for abnormal levels of hyperopia in children: a non-cycloplegic method with a hand held refractor. Br J Ophthalmol 1998;82:1260 – 4. 26. Cordonnier M, Kallay O. Non-cycloplegic screening for refractive errors in children with the hand-held autorefractor Retinomax: final results and comparison with non-cycloplegic photoscreening. Strabismus 2001;9:59 –70. 27. Rogers DL, Neely DE, Chapman JB, et al. Comparison of the MTI Photoscreener and the Welch-Allyn SureSight autorefractor in a tertiary care center. J AAPOS 2008;12:77– 82. 28. Salcido AA, Bradley J, Donahue SP. Predictive value of photoscreening and traditional screening of preschool children. J AAPOS 2005;9:114 –20. 29. Leone JF, Mitchell P, Morgan IG, et al. Use of visual acuity to screen for significant refractive errors in adolescents: is it reliable? Arch Ophthalmol 2010;128:894 –9. 30. Ohlsson J, Villarreal G, Abrahamsson M, et al. Screening merits of the Lang II, Frisby, Randot, Titmus, and TNO stereo tests. J AAPOS 2001;5:316 –22. 31. Fulton AB, Hansen RM, Petersen RA. The relation of myopia and astigmatism in developing eyes. Ophthalmology 1982;89: 298 –302. 32. Farbrother JE, Welsby JW, Guggenheim JA. Astigmatic axis is related to the level of spherical ametropia. Optom Vis Sci 2004;81:18 –26. 33. Murdoch IE, Morris SS, Cousens SN. People and eyes: statistical approaches in ophthalmology. Br J Ophthalmol 1998; 82:971–3. 34. Pan Y, Tarczy-Hornoch K, Cotter SA, et al, Multi-Ethnic Pediatric Eye Disease Study (MEPEDS) Group. Visual acuity norms in pre-school children: the Multi-Ethnic Pediatric Eye Disease Study. Optom Vis Sci 2009;86:607–12.

Footnotes and Financial Disclosures Originally received: May 15, 2012. Final revision: July 23, 2012. Accepted: August 3, 2012. Available online: November 22, 2012.

5

School of Nursing, College of Nursing, Kaohsiung Medical University, Kaohsiung, Taiwan.

6

Manuscript no. 2012-700.

1

Department of Ophthalmology, Kaohsiung Medical University Hospital, Kaohsiung Medical University, Kaohsiung, Taiwan.

2

Department of Ophthalmology, School of Medicine, College of Medicine, Kaohsiung Medical University, Kaohsiung, Taiwan.

3

Graduate Institute of Medicine, College of Medicine, Kaohsiung Medical University, Kaohsiung, Taiwan.

4

Department of Ophthalmology, Kaohsiung Municipal Hsiao-Kang Hospital, Kaohsiung Medical University, Kaohsiung, Taiwan.

276

Department of Kinesiology, Health, and Leisure Studies, College of Humanities and Social Sciences, National University of Kaohsiung, Kaohsiung, Taiwan. Financial Disclosure(s): The author(s) have no proprietary or commercial interest in any materials discussed in this article. Supported by a grant from the Chen C. W. Foundation for Blindness Prevention and Education (the funding organization had no role in the design or conduct of this research). Correspondence: Hwei-Zu Wang, MD, 100 Zih-You 1st Rd., Kaohsiung City 807, Taiwan. E-mail: [email protected].