1040-5488/13/9012-1479/0 VOL. 90, NO. 12, PP. 1479Y1485 OPTOMETRY AND VISION SCIENCE Copyright * 2013 American Academy of Optometry
ORIGINAL ARTICLE
Myopia Screening: Combining Visual Acuity and Noncycloplegic Autorefraction Yingyan Ma*, Xiangui He†, Haidong Zou‡, Lina Lu†, Xiaomei Qu‡, and Jianfeng Zhu‡
ABSTRACT Purpose. To explore the effectiveness of combining uncorrected visual acuity (UCVA) and noncycloplegic autorefraction (NCAR) for large-scale myopia screening in schoolchildren with a high prevalence of myopia. Methods. A total of 1687 children aged between 6 and 12 years, from five primary schools in the Baoshan district of Shanghai participated in the study. We measured UCVA and autorefraction before and after cycloplegia by a Topcon KR-8800. We drew receiver operating characteristic curves to achieve the best cutoff points and their corresponding sensitivities and specificities for the UCVA and NCAR, respectively. We then combined the UCVA and NCAR in serial order to explore the optimal criterion and its effectiveness. A specificity of 90% was set to compare the sensitivities among the three tests. The children were divided into three age groups (aged 6 to 7, 8 to 10, 11 to 12 years) to further examine this new method. Results. A total of 1639 children with an average age of 9.35 (SD, 1.6) years were finally included, among which 428 (26.11%) children were diagnosed as being myopic (spherical equivalent refraction (SER) less than or equal to j0.5 diopters [D]). For the UCVA, the cutoff point is 0.2 logarithm of the minimum angle of resolution (20/32), with a sensitivity and specificity of 63.6% and 94.0%, respectively. For NCAR, the cutoff point is SER less than or equal to j0.75 D, with a sensitivity and specificity of 88.6% and 86.1%, respectively. When UCVA is combined with NCAR, the best cutoff point is UCVA less than or equal to 0 logarithm of the minimum angle of resolution (20/20) and SER less than or equal to j0.75 D, with a sensitivity and specificity of 84.4% and 90.5%, respectively. At a specificity of 90%, the sensitivities are 63.55%, 78.50%, and 84.35%, respectively, for UCVA, NCAR, and the combination test. In all age groups, the combination test had the highest sensitivities among the three tests. Conclusions. In a population with a high prevalence of myopia, combining the UCVA and NCAR in serial order achieved higher sensitivity than either of the two tests alone, when specificity was set at 90%. (Optom Vis Sci 2013;90:1479Y1485) Key words: myopia, screening, visual acuity, autorefraction, cycloplegia
T
he prevalence of myopia is especially high for school-aged children in China.1Y5 In urban areas in Mainland China, the prevalence of myopia is reported to be 5.7% in 5-yearolds, 30.1% in 10-year-olds, and 78.4% in 15-year-olds.3 In rural areas of China, myopia is almost absent in 5-year-olds but increases to 38.8% in 15-year-olds.1 A relatively high prevalence
*MM † MPH ‡ MD Department of Ophthalmology, Shanghai First People’s Hospital, Affiliated Shanghai Jiaotong University, Shanghai, China (YM, HZ); Shanghai Eye Disease Prevention and Treatment Center, Shanghai, China (XH, HZ, LL, JZ); and Department of Ophthalmology, Shanghai Eye and ENT Hospital, Affiliated Fudan University, Shanghai, China (XQ).
of myopia is also observed in schoolchildren in Singapore, Hong Kong, and Taiwan.6Y8 The blurred vision caused by myopia prevents children from clearly seeing words on the blackboard, which may affect their studies and performance in school.9 Hence, screening for myopia in schoolchildren is beneficial because it helps inform parents of their children’s refractive status and promotes further examinations of the children’s refractive error. In addition, detecting myopia at an early stage allows for early treatment and preventive measures, which may control the progression of myopia.10,11 The most widely used method for myopia screening is the uncorrected visual acuity (UCVA) test. This test requires only a visual acuity chart, adequate illumination, and trained nurses or technicians. There have been several studies reporting the optimal visual acuity threshold for myopia screening in schoolchildren.12Y14
Optometry and Vision Science, Vol. 90, No. 12, December 2013
Copyright © American Academy of Optometry. Unauthorized reproduction of this article is prohibited.
1480 Myopia Screening by Acuity and Noncycloplegic AutorefractionVMa et al.
Tong and associates suggested a threshold of 0.26 logarithm of the minimum angle of resolution (logMAR) (Snellen 20/36) using the ETDRS (Early Treatment Diabetic Retinopathy Study) chart performed by optometrists. This had a sensitivity of 91.8% and a specificity of 87.5% for myopia screening. They also suggested a threshold of 0.30 logMAR (20/40) using a simplified visual screening chart, with testing performed by nurses. This had a sensitivity of 91% and a specificity of 83.5%.13 A myopia study conducted in Sydney, using a criterion of 6/9.5 (Snellen) by logMAR chart, had a sensitivity of 97.8% and a specificity of 91.7%.15 Another study in Northern Ireland obtained a threshold of 0.20 logMAR (20/32), with a sensitivity of 92% and a specificity of 91%.16 In reality, the accuracy of the test could be influenced by a number of factors, such as disturbance from noisy surroundings, illumination, comprehension ability of children, and different standards used by examiners. In addition, the UCVA test alone cannot differentiate between myopia and other types of refractive error and is therefore not a reliable measurement for myopia screening.15,16 The most commonly used visual acuity chart in China is the Chinese Standard Logarithmic Visual Acuity E Chart.17 Although the threshold of 0.10 logMAR (20/25) has been used in vision screening for many years, there is no validated reference regarding its accuracy for myopia screening. Cycloplegic autorefraction (CAR) is a reliable method for measuring refractive error in young children18Y20 and has been applied in optometric and ophthalmic research defining myopia.12,13,15,16 However, this method is not appropriate for largescale myopia screening because the process of cycloplegia requires ophthalmologists or optometrists, is time consuming to reach full cycloplegia, and cycloplegia causes blurred vision and carries the possibility of adverse effects. Conversely, noncycloplegic autorefraction (NCAR) is a simple procedure; it omits the process of cycloplegia and can be easily administered by trained personnel. Therefore, NCAR has been used in several studies for assessment of refractive error,21Y23 despite the fact that it can overestimate myopia and underestimate hyperopia in children, whose accommodative responses are active.24Y28 The combined use of NCAR and UCVA for astigmatism screening in Native American preschool children has been shown to improve specificity without reducing the sensitivity of the tests.29 Another recent study, screening for significant refractive errors in Taiwanese preschool children, also confirmed the advantage of combining these two methods.30 However, to our knowledge, there have been no studies on whether the combined methods can be effectively used for large-scale myopia screening in a school population with a high prevalence of myopia. The current study explores the criterion and the corresponding sensitivity and specificity of the combined use of NCAR and UCVA for myopia screening in Chinese school-aged children.
METHODS This study is a preliminary trial of a 3-year public health program to establish archives of refractive status of children in Shanghai. Five schools were chosen, by cluster random sampling, in the Baoshan district, a suburban area of Shanghai. The study was done between May and June 2010, during weekdays, while classes were in session. One ophthalmologist, three optometrists,
and two public health doctors conducted the study. All testing on each child was completed at one visit. Initially, the children underwent screening (UCVA and NCAR), followed by a detailed examination (subjective refraction, intraocular pressure measurement by noncontact tonometer, slit lamp examination, cycloplegia, and CAR). The ophthalmologist performed the slit lamp examination and was responsible for determining whether a child was suitable for cycloplegia. The optometrists conducted the NCAR, CAR, and subjective refraction. The public health doctors performed the UCVA and instilled cycloplegia. Schoolchildren from grades I to V, aged 6 to 12 years, participated in the study (n = 1687). Children who were allergic to tropicamide (n = 9), with a high intraocular pressure (either known before the test or discovered during the screening test) (n = 25), or who were uncooperative (n = 14) were also excluded. A total of 1639 schoolchildren (56.9% male), with a mean age of 9.35 years (SD, 1.60), completed all examinations and were included in the data analysis. This study was conducted according to the tenets of the Declaration of Helsinki and was approved by the institutional review board at Shanghai Eye Disease Prevention and Treatment Center. Written informed consent was received from parents of all the children who enrolled in the study. Uncorrected visual acuity was performed monocularly, in each eye, without optical aids at 5 m, using the Standard Logarithmic Visual Acuity E Chart.17 The chart has one optotype in 1.0 logMAR (20/200), two optotypes in line 0.92 logMAR (20/160) to line 0.82 logMAR (20/133), three optotypes in line 0.70 logMAR (20/100) to line 0.60 logMAR (20/80), four optotypes in line 0.48 logMAR (20/60) to line 0.40 logMAR (20/50), five optotypes in line 0.30 logMAR (20/40), six optotypes in line 0.20 logMAR (20/32), seven optotypes in line 0.10 logMAR (20/25), and eight optotypes in line 0 logMAR (20/20) to line j0.30 logMAR (20/10). The right eye was always examined first. The doctor started the examination at line 0 logMAR (20/20). If the child was able to recognize the optotype, the testing proceeded to the next smaller line; if the child failed, the testing proceeded to the next bigger line. If the child failed one optotype in line 1 logMAR (20/200) to 0.48 logMAR (20/60), two or more optotypes in line logMAR 0.40 (20/50) to 0 logMAR (20/20), or three or more optotypes in line j0.10 logMAR (20/16) to j0.30 logMAR (20/10), the visual acuity was recorded as the line above. The child was required to tell each optotype within 5 seconds; if a longer time was taken, the doctor would proceed to another optotype. The child was not allowed to squeeze or squint during the examination. A table-mounted KR-8800 (Topcon, Tokyo, Japan), a commonly used autorefractor in Asia, with a measurement range of j20 to +20 diopters (D), was used to obtain autorefraction data from each eye beginning with the right eye, before and after cycloplegia. Three consecutive readings of sphere, cylinder, and axis of each eye were taken, and the average was used for analysis. The autorefractometer was calibrated before each examination. Cycloplegia was induced by instillation of 0.5% tropicamide, the most widely used cycloplegic agent for myopia screening in China. One drop was administered every 5 minutes for a total of 5 drops. Although tropicamide inhibits accommodation less effectively than cyclopentolate,31 many studies have reported it to be an effective cycloplegic agent for myopic children in dark
Optometry and Vision Science, Vol. 90, No. 12, December 2013
Copyright © American Academy of Optometry. Unauthorized reproduction of this article is prohibited.
Myopia Screening by Acuity and Noncycloplegic AutorefractionVMa et al.
irises.32Y34 Cycloplegia was considered complete when the pupil dilated to 6 mm or greater and a light reflex was absent. Autorefraction measurements were carried out at an average of 40 minutes after the first drop of tropicamide, if adequate cycloplegia was obtained. If a child did not reach full dilation at the time of measurement, one or two additional drops were given to get full dilation. Because unaided visual acuity, NCAR, and cycloplegic autorefraction results were closely related between the right and the left eyes (Spearman correlations were 0.745, 0.817, and 0.845 for UCVA, NCAR, and CAR, respectively; values of p G 0.001), the right eye was chosen for data analysis. In this study, myopia was defined as a spherical equivalent refraction (sphere + 0.5 cylinder, SER) of at least j0.5 D. Cycloplegic autorefraction was used as the reference standard because this method of examination has been proven to be reliable and stable for detecting refractive error in young children.18,19 Receiver-operating characteristic (ROC) curves were constructed to obtain the best cutoff value, area under the curve (AUC), and Youden index for each test (UCVA and NCAR). Sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and the 95% confidence intervals (95% CIs) were also calculated for every referral threshold (MedCalc 11.4.2.0). A specificity of 90% was set for the purpose of screening,27,35 and the criteria with the highest sensitivity were selected as optimal. We then combined UCVA and NCAR in serial order (children were referred if they failed both tests) to explore the optimal criteria and their corresponding sensitivity, specificity, and Youden Index for myopia screening. Children were divided into three age groups as follows: 6 to 7 years, 8 to 10 years, and 11 to 12 years to further explore the effectiveness of this combination. A paired t test was used to compare the difference in SER before and after cycloplegia.
1481
corresponding sensitivities, specificities, predictive values, and Youden Indexes are listed in Table 1. Fig. 2 shows the ROC curve for the NCAR test. The AUC was 0.940 (95% CI, 0.928 to 0.951), larger than the AUC for the UCVA test (p G 0.0001), indicating a greater ability of the NCAR test to discriminate myopia compared with the UCVA. The best cutoff point was an SER of j0.75 D, with a sensitivity of 88.55% and a specificity of 86.13% (Table 2). This criterion can be accepted when sensitivity and specificity are regarded as equally important. To reduce the cost of unnecessary referral after screening, a higher specificity was preferred in this study because the screening was applied to a population on a large scale.36 In this study, we selected a specificity of 90% as the standard for the screening tests, which is commonly used for mass screening.27,35 Using this standard, the criterion for SER of j1.0 D or less had a sensitivity of 78.50% and a specificity of 91.91%. Performing two diagnostic tests in serial order can theoretically improve specificity but can reduce sensitivity. Namely, the specificity of the combined test is higher than that of either test alone, but the sensitivity of the combined test is lower than that of either test.37 Combining UCVA with NCAR has been reported to have increased the specificity without significantly dropping the sensitivity in preschool children.29,30 Therefore, to achieve more satisfactory sensitivity results along with the required specificity, we tried different combinations of UCVA with NCAR tests in serial order. The best combination was UCVA less than or equal to 0 logMAR (20/20) and NCAR less than or equal to j0.75 D, which had a sensitivity of 84.35% and a specificity of 90.50%. Compared with either using UCVA less than or equal to 0 logMAR (20/20) or NCAR less than or equal to j0.75 D as referral criteria, the combined test increased specificity without
RESULTS Of the 1639 eyes included in the analysis, 428 (26.11%) eyes were diagnosed as being myopic by cycloplegic autorefraction. The median UCVA was 0 logMAR (20/20) for the study eyes (N = 1639). There were 256 children aged 6 to 7 years, 931 children aged 8 to 10 years, and 452 children aged 11 to 12 years. Using the data from the right eye, the prevalence rate of myopia (SER less than or equal to j0.5 D) was 8.98% for ages 6 to 7 years, 22.23% for ages 8 to 10 years, and 43.81% for ages 11 to 12 years. The mean SER before cycloplegia was j0.51 T 1.20 D for all children and 0.00 T1.31 D after cycloplegia. The mean SER was an average of 0.51 D more myopic when measured with NCAR than with cycloplegia autorefraction (p G 0.001). Fig. 1 shows the ROC curve for UCVA. The area under the ROC curve was 0.851 (95% CI, 0.833 to 0.868). The cutoff point was a visual acuity of 0.20 logMAR (20/32), with the corresponding sensitivity of 63.55% and specificity of 93.97%. With a UCVA of 0.10 logMAR (20/25), the most commonly used criterion for vision screening in China, the sensitivity is 80.14% and the specificity is 76.71%. The relatively low specificity means that the UCVA test is likely to incorrectly refer 23.29% of children with normal vision. Other referral criteria and their
FIGURE 1. Receiver operating characteristic curve of UCVA for myopia screening in schoolchildren. The points for UCVA of 0.2 logMAR (Snellen 20/32) and 0.1 logMAR (Snellen 20/25) were labeled. LogMAR, logarithm of the minimum angle of resolution; UCVA, uncorrected visual acuity.
Optometry and Vision Science, Vol. 90, No. 12, December 2013
Copyright © American Academy of Optometry. Unauthorized reproduction of this article is prohibited.
1482 Myopia Screening by Acuity and Noncycloplegic AutorefractionVMa et al. TABLE 1.
Sensitivity, specificity, PPV, NPV, and Youden Index of UCVA for schoolchildren (N = 1639) LogMAR (Snellen) 0.40 (20/50) 0.30 (20/40) 0.20 (20/32) 0.10 (20/25) 0 (20/20) j0.10 (20/16)
Sensitivity
95% CI
Specificity
95% CI
PPV
95% CI
NPV
95% CI
Youden Index
30.84 47.43 63.55 80.14 92.29 98.83
26.5Y35.5 42.6Y52.3 58.8Y68.1 76.0Y83.8 89.3Y94.6 97.3Y99.6
99.01 97.52 93.97 76.71 38.23 6.94
98.3Y99.5 96.5Y98.3 92.5Y95.2 74.2Y79.1 35.5Y41.0 5.6Y8.5
91.7 87.1 78.8 54.9 34.6 27.3
85.9Y95.6 82.1Y91.1 74.1Y83.0 50.9Y58.8 31.8Y37.4 25.1Y29.6
80.2 84.0 87.9 91.6 93.3 94.4
78.1Y82.2 82.0Y85.9 86.0Y89.7 89.7Y93.3 90.8Y95.4 87.4Y98.2
0.2985 0.4495 0.5752 0.5685 0.3052 0.0577
LogMAR, logarithm of the minimum angle of resolution; NPV, negative predictive value; PPV, positive predictive value; UCVA, uncorrected visual acuity.
substantially decreasing sensitivity. In addition, at a specificity of 90%, the sensitivities are 63.55%, 78.50%, and 84.35%, respectively for UCVA, NCAR, and combination of the two (Table 3). Moreover, the Youden Index, which describes the accuracy of a test, was higher for the combination test than for either the UCVA test or the NCAR test alone, suggesting that the combination test is better than the individual tests for diagnosing myopia. We then conducted further analyses in the three different age groups (Table 3). When the specificity was set to 90%, sensitivity was always the lowest for the UCVA test alone and the highest for the combination of the UCVA and NCAR tests in all age groups. The Youden Index of the combination test was also the highest for all age groups, indicating that the accuracy of the combination test is better than that of the UCVA or NCAR test alone. The best criteria for the combination test were a little different among the three age groups. Although applying the combination of UCVA less than or equal to 0 logMAR (20/20) and NCAR less than or equal to j0.75 D can achieve a specificity close to or more than 90% in all age groups, the sensitivity was not always the highest. For children aged 6 to 7 years, the combined use of UCVA less than or equal to 0.1 logMAR (20/25) and NCAR less than or equal to j0.5 D obtained a sensitivity of 91.30%, which was higher than that for the combination of UCVA less than or equal to 0 logMAR (20/20) and NCAR less than or equal to j0.75 D. This result suggests that, to obtain the optimal effect, different criteria need to be established for different age groups when applying the combination test for myopia screening.
the UCVA may be better than that of the Chinese Standard Logarithmic Visual Acuity E Chart likely because of the use of letter-by-letter rather than line-by-line scoring.13 Although the Topcon KR-8800 has been commonly used for refractive measurements in Asia,38,39 the accuracy of this autorefractor has not been reported. In this study, at the optimal cutoff point of j0.75 D (SER less than or equal to j0.75 D), the NCAR test using the Topcon KR-8800 had a sensitivity of 88.55% and a specificity of 86.13%. These results are similar to those of the NCAR test that used the Topcon KR-8100 when screening for significant refractive error in Taiwanese preschool children.30 Compared with other types of autorefractors, the NCAR by the Topcon KR-8800 had a relatively high sensitivity and specificity when used for myopia screening in school-aged children, albeit not the highest.40,41 Moreover, our study shows that the accuracy of the NCAR test is better than that of the UCVA test for detecting myopia, which is demonstrated by the larger AUC of the former. This phenomenon has also been
DISCUSSION In our study, the optimal criterion for UCVA using the Chinese Standard Logarithmic Visual Acuity E Chart was 0.2 logMAR (20/32), with a sensitivity of 63.55% and a specificity of 93.97%. Compared with the visual acuity of 0.1 logMAR (20/25), the most frequently used criterion for vision screening in China, it has a satisfactory specificity, however, a lower sensitivity. The prevalence of myopia is high among schoolchildren in China, and a test with a sensitivity of 63.55% would not detect 36.45% of myopic children. The accuracy of the UCVA test in the study seems to be lower than that reported in other studies.12,15,16 This discrepancy might result from the different standards for defining myopia and different methods to record visual acuity. For instance, sometimes an SER of j1.0 D or higher was used to define myopia.16 The accuracy of the EDTRS chart for measuring
FIGURE 2. Receiver operating characteristic curve of NCAR for myopia screening in schoolchildren. The points for SER of j0.5 D, j0.75 D, j1.0 D, and j1.25 D were labeled. NCAR, noncycloplegic autorefraction; SER, spherical equivalent refraction.
Optometry and Vision Science, Vol. 90, No. 12, December 2013
Copyright © American Academy of Optometry. Unauthorized reproduction of this article is prohibited.
Myopia Screening by Acuity and Noncycloplegic AutorefractionVMa et al.
1483
TABLE 2.
Sensitivity, specificity, PPV, NPV, their 95% CI, and Youden Index of NCAR for schoolchildren (N = 1639) SER, D
Sensitivity
95% CI
Specificity
95% CI
PPV
95% CI
NPV
95% CI
Youden Index
j1.5 j1.25 j1 j0.75 j0.5 j0.25
52.10 66.36 78.50 88.55 95.56 98.83
47.3Y56.9 61.7Y70.8 74.3Y82.3 85.1Y91.4 93.2Y97.3 97.3Y99.6
97.69 95.87 91.91 86.13 75.64 57.64
96.7Y98.5 94.6Y96.9 90.2Y93.4 84.1Y88.0 73.1Y78.0 54.8Y60.4
88.8 85.0 77.4 69.3 58.1 45.2
84.3Y92.5 80.7Y88.7 73.2Y81.3 65.2Y73.1 54.4Y61.8 42.0Y48.4
85.2 89.0 92.4 95.5 98.0 99.3
83.3Y87.1 87.1Y90.6 90.7Y93.8 94.1Y96.7 96.8Y98.8 98.3Y99.8
0.4979 0.6223 0.7041 0.7468 0.7120 0.5647
NCAR, noncycloplegic autorefraction; NPV, negative predictive value; PPV, positive predictive value; SER, spherical equivalent refraction.
observed in studies that focused on screening for refractive error in preschool children.29,30 Attempts to measure refractive status without cycloplegia can lead to significant errors, such as an overestimation of myopia because of the strong accommodative reserve, especially in children younger than 12 years.26 In our study, NCAR overestimates the extent of myopia by, on average, approximately 0.50 D. This disadvantage of NCAR hampers its widespread application in myopia screening in school-aged children. Whether satisfactory screening efficacy can be obtained through the combined use of UCVA and NCAR for myopia screening in schoolchildren has not yet been reported in previous research. Theoretically, with a high sensitivity but a low specificity, using the criterion of UCVA less than or equal to 0 logMAR (20/20) alone cannot discriminate hyperopic and astigmatic eyes whose UCVA were 0 logMAR (20/20) or less. However, combining UCVA with NCAR may eliminate these false-positive conditions
to some extent. Meanwhile, using a single NCAR test can overestimate myopic refraction in school-aged children, mistakenly diagnosing as myopic those children who are emmetropic or hyperopic. However, those people usually have a UCVA of 0 logMAR (20/20) or more, which can be distinguished by combining NCAR with the UCVA test. Combining the two methods in serial order increased the specificity without extensively decreasing the sensitivity in schoolaged children; these results are similar to those found in previous studies in preschool children.29,30 In addition, at a specificity of 90%, the sensitivity for the combination test was always the highest among the three tests in all age groups. Hence, the combination test should not only reduce the cost of unnecessary referrals to ophthalmic departments but also detect more myopic children than either of the two tests alone. In the present study of 1639 subjects, compared with the UCVA test of 0.8 (20/25), the most widely used method for myopia screening in China,21
TABLE 3.
Sensitivity, specificity, and criteria for UCVA, NCAR, and combined use of UCVA and NCAR when specificity is set to 90% (N = 1639) Sensitivity, %
Specificity, %
Youden Index
Criteria
All (N = 1639) UCVA NCAR Combination*
63.55 78.50 84.35
93.97 91.91 90.50
0.575 0.704 0.749
0.2 logMAR (Snellen 20/32) SER e j1.0 D UCVA e 0 logMAR (20/20) and SER e j0.75 D
6Y7 y (n = 256) UCVA NCAR Combination 1* Combination 2*
69.49 86.96 86.96 91.30
94.15 91.42 93.99 92.37
0.636 0.784 0.810 0.837
0.2 logMAR (20/32) SER e j0.75 D UCVA e 0 logMAR (20/20) and SER e j0.75 D UCVA e 0.1 logMAR (20/25) and SER e j0.5 D
8Y10 y (n = 931) UCVA NCAR Combination 1* Combination 2*
67.15 77.78 80.68 85.02
93.78 91.71 93.37 89.23
0.609 0.695 0.741 0.743
0.2 logMAR (20/32) SER e j1.0 D UCVA e 0.1 logMAR (20/25) and SER e j0.5 D UCVA e 0 logMAR (20/20) and SER e j0.75 D
11Y12 y (n = 452) UCVA NCAR Combination*
59.09 78.79 83.33
94.49 90.94 90.94
0.536 0.697 0.743
0.2 logMAR (20/32) SER e j1.0 D UCVA e 0 logMAR (20/20) and SER e j0.75 D
*Combination means combined use of UCVA and NCAR in serial order (children are referred if they failed both tests). LogMAR, logarithm of the minimum angle of resolution; NCAR, noncycloplegic autorefraction; SER, spherical equivalent refraction; UCVA, uncorrected visual acuity. Optometry and Vision Science, Vol. 90, No. 12, December 2013
Copyright © American Academy of Optometry. Unauthorized reproduction of this article is prohibited.
1484 Myopia Screening by Acuity and Noncycloplegic AutorefractionVMa et al.
the combination test resulted in 18 more true-positive subjects (true-positive number: UCVA, n = 343; combination test, n = 361), and 167 fewer false-positive subjects with myopia (falsepositive number: UCVA, n = 282; combination test, n = 155). The cost for an autorefractor is approximately 80,000 RMB (12,878 US dollars) and the cost for an unnecessary referral to the ophthalmology department is estimated to be 100 RMB (16 US dollars) per child, according to the average outpatient service charge in the ophthalmic department of Shanghai. Therefore, with 7851 subjects included, the cost for the autorefractor could be compensated, rendering the combination test to be a cost-effective measurement. Moreover, without cycloplegia, the procedure is simple and time saving, and does not require additional personnel to administer the cycloplegic agents. Like most studies, there are some limitations in our study. First, the combination test cannot discriminate children with pseudomyopia who might have UCVA less than or equal to 0 logMAR (20/20) and NCAR less than or equal to j0.75 D but SER of less than j0.5 D after cycloplegia. Second, some children with true myopia might have NCAR greater than j0.75 D and UCVA greater than 0 logMAR (20/20) but are left undetected by the combination criterion. Moreover, as sensitivity and specificity results change with prevalence rates of diseases,42 the results of the combined test might not be effectively applied in populations with distinctly different prevalence rates of myopia. Last, but not least, the best criterion for screening myopia by the combination test cannot be arbitrarily determined in the present study. As shown in the results, the optimal criteria for screening myopia were not consistent among the different age groups. Different criteria may be established according to different age groups. In conclusion, the combined use of the UCVA and NCAR in serial order achieved higher sensitivity than either of the two tests alone, when the specificity was set at 90%. This new method can be applied to large-scale myopia screening in Chinese schoolchildren aged 6 to 12 years.
ACKNOWLEDGMENTS This research is supported by Funding of Scientific Research for young researchers by Shanghai Health Bureau (No. 2010Y113). No commercial interest or potential conflict of interest exists for any author. The first two authors contributed equally to this work and are considered co-first authors. Received March 30, 2013; accepted August 30, 2013.
REFERENCES 1. Zhao J, Pan X, Sui R, Munoz SR, Sperduto RD, Ellwein LB. Refractive Error Study in Children: results from Shunyi District, China. Am J Ophthalmol 2000;129:427Y35. 2. Congdon N, Wang Y, Song Y, Choi K, Zhang M, Zhou Z, Xie Z, Li L, Liu X, Sharma A, Wu B, Lam DS. Visual disability, visual function, and myopia among rural chinese secondary school children: the Xichang Pediatric Refractive Error Study (X-PRES)Vreport 1. Invest Ophthalmol Vis Sci 2008;49:2888Y94. 3. He M, Zeng J, Liu Y, Xu J, Pokharel GP, Ellwein LB. Refractive error and visual impairment in urban children in southern China. Invest Ophthalmol Vis Sci 2004;45:793Y9.
4. Pan CW, Ramamurthy D, Saw SM. Worldwide prevalence and risk factors for myopia. Ophthalmic Physiol Opt 2012;32:3Y16. 5. He M, Zheng Y, Xiang F. Prevalence of myopia in urban and rural children in mainland China. Optom Vis Sci 2009;86:40Y4. 6. Edwards MH, Lam CS. The epidemiology of myopia in Hong Kong. Ann Acad Med Singapore 2004;33:34Y8. 7. Saw SM, Carkeet A, Chia KS, Stone RA, Tan DT. Component dependent risk factors for ocular parameters in Singapore Chinese children. Ophthalmology 2002;109:2065Y71. 8. Lin LL, Shih YF, Hsiao CK, Chen CJ. Prevalence of myopia in Taiwanese schoolchildren: 1983 to 2000. Ann Acad Med Singapore 2004;33:27Y33. 9. Mathers M, Keyes M, Wright M. A review of the evidence on the effectiveness of children’s vision screening. Child Care Health Dev 2010;36:756Y80. 10. Morgan IG, Ohno-Matsui K, Saw SM. Myopia. Lancet 2012;379: 1739Y48. 11. Walline JJ, Lindsley K, Vedula SS, Cotter SA, Mutti DO, Twelker JD. Interventions to slow progression of myopia in children. Cochrane Database Syst Rev 2011:CD004916. 12. Tong L, Saw SM, Tan D, Chia KS, Chan WY, Carkeet A, Chua WH, Hong CY. Sensitivity and specificity of visual acuity screening for refractive errors in school children. Optom Vis Sci 2002; 79:650Y7. 13. Tong L, Saw SM, Chan ES, Yap M, Lee HY, Kwang YP, Tan D. Screening for myopia and refractive errors using LogMAR visual acuity by optometrists and a simplified visual acuity chart by nurses. Optom Vis Sci 2004;81:684Y91. 14. Luo HD, Gazzard G, Liang Y, Shankar A, Tan DT, Saw SM. Defining myopia using refractive error and uncorrected logMAR visual acuity 90.3 from 1334 Singapore school children ages 7Y9 years. Br J Ophthalmol 2006;90:362Y6. 15. Leone JF, Mitchell P, Morgan IG, Kifley A, Rose KA. Use of visual acuity to screen for significant refractive errors in adolescents: is it reliable? Arch Ophthalmol 2010;128:894Y9. 16. O’Donoghue L, Rudnicka AR, McClelland JF, Logan NS, Saunders KJ. Visual acuity measures do not reliably detect childhood refractive errorYan epidemiological study. PLoS One 2012;7:e34441. 17. National Standard of People’s Republic of China. Standard Logarithmic Visual Acuity E Chart (GB 11533-1989). Beijing, China: China Standard Press; 1990. 18. Choong YF, Chen AH, Goh PP. A comparison of autorefraction and subjective refraction with and without cycloplegia in primary school children. Am J Ophthalmol 2006;142:68Y74. 19. Harvey EM, Miller JM, Dobson V, Tyszko R, Davis AL. Measurement of refractive error in Native American preschoolers: validity and reproducibility of autorefraction. Optom Vis Sci 2000; 77:140Y9. 20. Pedamallu SB, Reddy KS, Pedamallu R, Pedamallu CS. Reliability of cycloplegic autorefractor measurements to determine spherical and astigmatic refractive errors in young children. Internet J Ophthalmol Visual Sci 2008;6(1):doi: 10.5580/1663. 21. Fotouhi A, Hashemi H, Khabazkhoob M, Mohammad K. The prevalence of refractive errors among schoolchildren in Dezful, Iran. Br J Ophthalmol 2007;91:287Y92. 22. Williams C, Miller L, Northstone K, Sparrow JM. The use of noncycloplegic autorefraction data in general studies of children’s development. Br J Ophthalmol 2008;92:723Y4. 23. Fotouhi A, Morgan IG, Iribarren R, Khabazkhoob M, Hashemi H. Validity of noncycloplegic refraction in the assessment of refractive errors: the Tehran Eye Study. Acta Ophthalmol 2012;90:380Y6.
Optometry and Vision Science, Vol. 90, No. 12, December 2013
Copyright © American Academy of Optometry. Unauthorized reproduction of this article is prohibited.
Myopia Screening by Acuity and Noncycloplegic AutorefractionVMa et al. 24. Harvey EM, Miller JM, Dobson V, Tyszko R, Davis AL. Measurement of refractive error in Native American preschoolers: validity and reproducibility of autorefraction. Optom Vis Sci 2000; 77:140Y9. 25. Zhao J, Mao J, Luo R, Li F, Pokharel GP, Ellwein LB. Accuracy of noncycloplegic autorefraction in school-age children in China. Optom Vis Sci 2004;81:49Y55. 26. Fotedar R, Rochtchina E, Morgan I, Wang JJ, Mitchell P, Rose KA. Necessity of cycloplegia for assessing refractive error in 12-year-old children: a population-based study. Am J Ophthalmol 2007;144:307Y9. 27. Schmidt P, Maguire M, Dobson V, Quinn G, Ciner E, Cyert L, Kulp MT, Moore B, Orel-Bixler D, Redford M, Ying GS. Comparison of preschool vision screening tests as administered by licensed eye care professionals in the Vision In Preschoolers Study. Vision in Preschoolers (VIP) Study Group. Ophthalmology 2004;111:637Y50. 28. Jorge J, Queiros A, Gonzalez-Meijome J, Fernandes P, Almeida JB, Parafita MA. The influence of cycloplegia in objective refraction. Ophthalmic Physiol Opt 2005;25:340Y5. 29. Miller JM, Harvey EM, Dobson V. Visual acuity screening versus noncycloplegic autorefraction screening for astigmatism in Native American preschool children. J AAPOS 1999;3:160Y5. 30. Lai YH, Tseng HY, Hsu HT, Chang SJ, Wang HZ. Uncorrected visual acuity and noncycloplegic autorefraction predict significant refractive errors in Taiwanese preschool children. Ophthalmology 2013;120:271Y6. 31. Egashira SM, Kish LL, Twelker JD, Mutti DO, Zadnik K, Adams AJ. Comparison of cyclopentolate versus tropicamide cycloplegia in children. Optom Vis Sci 1993;70:1019Y26. 32. Hamasaki I, Hasebe S, Kimura S, Miyata M, Ohtsuki H. Cycloplegic effect of 0.5% tropicamide and 0.5% phenylephrine mixed eye drops: objective assessment in Japanese schoolchildren with myopia. Jpn J Ophthalmol 2007;51:111Y5. 33. Fan DS, Rao SK, Ng JS, Yu CB, Lam DS. Comparative study on the safety and efficacy of different cycloplegic agents in children with darkly pigmented irides. Clin Experiment Ophthalmol 2004; 32:462Y7.
1485
34. Chen X, Lin Z, Lai X, Fang B, Li C, Li L. An objective investigation on the cycloplegic effects of tropicamide. China J Optom Ophthalmol 2008;10:135Y8. 35. Vision in Preschoolers (VIP) Study Group. Preschool vision screening tests administered by nurse screeners compared with lay screeners in the vision in preschoolers study. Invest Ophthalmol Vis Sci 2005; 46:2639Y48. 36. Wilson JM, Jungner G. Principles and Practice of Screening for Disease. World Health Organization Public Health Paper No. 34. Geneva, Switzerland: World Health Organization; 1968. 37. Weinstein S, Obuchowski NA, Lieber ML. Clinical evaluation of diagnostic tests. AJR Am J Roentgenol 2005;184:14Y9. 38. Fotouhi A, Hashemi H, Yekta AA, Mohammad K, Khoob MK. Characteristics of astigmatism in a population of schoolchildren, Dezful, Iran. Optom Vis Sci 2011;88:1054Y9. 39. Shen P, Zheng Y, Ding X, Liu B, Congdon N, Morgan I, He M. Biometric measurements in highly myopic eyes. J Cataract Refract Surg 2013;39:180Y7. 40. 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;9:59Y70. 41. Rogers DL, Neely DE, Chapman JB, Plager DA, Sprunger DT, Sondhi N, Roberts GJ, Ofner S. Comparison of the MTI Photoscreener and the Welch-Allyn SureSight autorefractor in a tertiary care center. J AAPOS 2008;12:77Y82. 42. Brenner H, Gefeller O. Variation of sensitivity, specificity, likelihood ratios and predictive values with disease prevalence. Stat Med 1997; 16:981Y91.
Jianfeng Zhu No. 380 Kangding Rd, Jingan District Shanghai Eye Disease Prevention and Treatment Center Shanghai 200040 China e-mail: [email protected]
Optometry and Vision Science, Vol. 90, No. 12, December 2013
Copyright © American Academy of Optometry. Unauthorized reproduction of this article is prohibited.
ORIGINAL ARTICLE
Myopia Screening: Combining Visual Acuity and Noncycloplegic Autorefraction Yingyan Ma*, Xiangui He†, Haidong Zou‡, Lina Lu†, Xiaomei Qu‡, and Jianfeng Zhu‡
ABSTRACT Purpose. To explore the effectiveness of combining uncorrected visual acuity (UCVA) and noncycloplegic autorefraction (NCAR) for large-scale myopia screening in schoolchildren with a high prevalence of myopia. Methods. A total of 1687 children aged between 6 and 12 years, from five primary schools in the Baoshan district of Shanghai participated in the study. We measured UCVA and autorefraction before and after cycloplegia by a Topcon KR-8800. We drew receiver operating characteristic curves to achieve the best cutoff points and their corresponding sensitivities and specificities for the UCVA and NCAR, respectively. We then combined the UCVA and NCAR in serial order to explore the optimal criterion and its effectiveness. A specificity of 90% was set to compare the sensitivities among the three tests. The children were divided into three age groups (aged 6 to 7, 8 to 10, 11 to 12 years) to further examine this new method. Results. A total of 1639 children with an average age of 9.35 (SD, 1.6) years were finally included, among which 428 (26.11%) children were diagnosed as being myopic (spherical equivalent refraction (SER) less than or equal to j0.5 diopters [D]). For the UCVA, the cutoff point is 0.2 logarithm of the minimum angle of resolution (20/32), with a sensitivity and specificity of 63.6% and 94.0%, respectively. For NCAR, the cutoff point is SER less than or equal to j0.75 D, with a sensitivity and specificity of 88.6% and 86.1%, respectively. When UCVA is combined with NCAR, the best cutoff point is UCVA less than or equal to 0 logarithm of the minimum angle of resolution (20/20) and SER less than or equal to j0.75 D, with a sensitivity and specificity of 84.4% and 90.5%, respectively. At a specificity of 90%, the sensitivities are 63.55%, 78.50%, and 84.35%, respectively, for UCVA, NCAR, and the combination test. In all age groups, the combination test had the highest sensitivities among the three tests. Conclusions. In a population with a high prevalence of myopia, combining the UCVA and NCAR in serial order achieved higher sensitivity than either of the two tests alone, when specificity was set at 90%. (Optom Vis Sci 2013;90:1479Y1485) Key words: myopia, screening, visual acuity, autorefraction, cycloplegia
T
he prevalence of myopia is especially high for school-aged children in China.1Y5 In urban areas in Mainland China, the prevalence of myopia is reported to be 5.7% in 5-yearolds, 30.1% in 10-year-olds, and 78.4% in 15-year-olds.3 In rural areas of China, myopia is almost absent in 5-year-olds but increases to 38.8% in 15-year-olds.1 A relatively high prevalence
*MM † MPH ‡ MD Department of Ophthalmology, Shanghai First People’s Hospital, Affiliated Shanghai Jiaotong University, Shanghai, China (YM, HZ); Shanghai Eye Disease Prevention and Treatment Center, Shanghai, China (XH, HZ, LL, JZ); and Department of Ophthalmology, Shanghai Eye and ENT Hospital, Affiliated Fudan University, Shanghai, China (XQ).
of myopia is also observed in schoolchildren in Singapore, Hong Kong, and Taiwan.6Y8 The blurred vision caused by myopia prevents children from clearly seeing words on the blackboard, which may affect their studies and performance in school.9 Hence, screening for myopia in schoolchildren is beneficial because it helps inform parents of their children’s refractive status and promotes further examinations of the children’s refractive error. In addition, detecting myopia at an early stage allows for early treatment and preventive measures, which may control the progression of myopia.10,11 The most widely used method for myopia screening is the uncorrected visual acuity (UCVA) test. This test requires only a visual acuity chart, adequate illumination, and trained nurses or technicians. There have been several studies reporting the optimal visual acuity threshold for myopia screening in schoolchildren.12Y14
Optometry and Vision Science, Vol. 90, No. 12, December 2013
Copyright © American Academy of Optometry. Unauthorized reproduction of this article is prohibited.
1480 Myopia Screening by Acuity and Noncycloplegic AutorefractionVMa et al.
Tong and associates suggested a threshold of 0.26 logarithm of the minimum angle of resolution (logMAR) (Snellen 20/36) using the ETDRS (Early Treatment Diabetic Retinopathy Study) chart performed by optometrists. This had a sensitivity of 91.8% and a specificity of 87.5% for myopia screening. They also suggested a threshold of 0.30 logMAR (20/40) using a simplified visual screening chart, with testing performed by nurses. This had a sensitivity of 91% and a specificity of 83.5%.13 A myopia study conducted in Sydney, using a criterion of 6/9.5 (Snellen) by logMAR chart, had a sensitivity of 97.8% and a specificity of 91.7%.15 Another study in Northern Ireland obtained a threshold of 0.20 logMAR (20/32), with a sensitivity of 92% and a specificity of 91%.16 In reality, the accuracy of the test could be influenced by a number of factors, such as disturbance from noisy surroundings, illumination, comprehension ability of children, and different standards used by examiners. In addition, the UCVA test alone cannot differentiate between myopia and other types of refractive error and is therefore not a reliable measurement for myopia screening.15,16 The most commonly used visual acuity chart in China is the Chinese Standard Logarithmic Visual Acuity E Chart.17 Although the threshold of 0.10 logMAR (20/25) has been used in vision screening for many years, there is no validated reference regarding its accuracy for myopia screening. Cycloplegic autorefraction (CAR) is a reliable method for measuring refractive error in young children18Y20 and has been applied in optometric and ophthalmic research defining myopia.12,13,15,16 However, this method is not appropriate for largescale myopia screening because the process of cycloplegia requires ophthalmologists or optometrists, is time consuming to reach full cycloplegia, and cycloplegia causes blurred vision and carries the possibility of adverse effects. Conversely, noncycloplegic autorefraction (NCAR) is a simple procedure; it omits the process of cycloplegia and can be easily administered by trained personnel. Therefore, NCAR has been used in several studies for assessment of refractive error,21Y23 despite the fact that it can overestimate myopia and underestimate hyperopia in children, whose accommodative responses are active.24Y28 The combined use of NCAR and UCVA for astigmatism screening in Native American preschool children has been shown to improve specificity without reducing the sensitivity of the tests.29 Another recent study, screening for significant refractive errors in Taiwanese preschool children, also confirmed the advantage of combining these two methods.30 However, to our knowledge, there have been no studies on whether the combined methods can be effectively used for large-scale myopia screening in a school population with a high prevalence of myopia. The current study explores the criterion and the corresponding sensitivity and specificity of the combined use of NCAR and UCVA for myopia screening in Chinese school-aged children.
METHODS This study is a preliminary trial of a 3-year public health program to establish archives of refractive status of children in Shanghai. Five schools were chosen, by cluster random sampling, in the Baoshan district, a suburban area of Shanghai. The study was done between May and June 2010, during weekdays, while classes were in session. One ophthalmologist, three optometrists,
and two public health doctors conducted the study. All testing on each child was completed at one visit. Initially, the children underwent screening (UCVA and NCAR), followed by a detailed examination (subjective refraction, intraocular pressure measurement by noncontact tonometer, slit lamp examination, cycloplegia, and CAR). The ophthalmologist performed the slit lamp examination and was responsible for determining whether a child was suitable for cycloplegia. The optometrists conducted the NCAR, CAR, and subjective refraction. The public health doctors performed the UCVA and instilled cycloplegia. Schoolchildren from grades I to V, aged 6 to 12 years, participated in the study (n = 1687). Children who were allergic to tropicamide (n = 9), with a high intraocular pressure (either known before the test or discovered during the screening test) (n = 25), or who were uncooperative (n = 14) were also excluded. A total of 1639 schoolchildren (56.9% male), with a mean age of 9.35 years (SD, 1.60), completed all examinations and were included in the data analysis. This study was conducted according to the tenets of the Declaration of Helsinki and was approved by the institutional review board at Shanghai Eye Disease Prevention and Treatment Center. Written informed consent was received from parents of all the children who enrolled in the study. Uncorrected visual acuity was performed monocularly, in each eye, without optical aids at 5 m, using the Standard Logarithmic Visual Acuity E Chart.17 The chart has one optotype in 1.0 logMAR (20/200), two optotypes in line 0.92 logMAR (20/160) to line 0.82 logMAR (20/133), three optotypes in line 0.70 logMAR (20/100) to line 0.60 logMAR (20/80), four optotypes in line 0.48 logMAR (20/60) to line 0.40 logMAR (20/50), five optotypes in line 0.30 logMAR (20/40), six optotypes in line 0.20 logMAR (20/32), seven optotypes in line 0.10 logMAR (20/25), and eight optotypes in line 0 logMAR (20/20) to line j0.30 logMAR (20/10). The right eye was always examined first. The doctor started the examination at line 0 logMAR (20/20). If the child was able to recognize the optotype, the testing proceeded to the next smaller line; if the child failed, the testing proceeded to the next bigger line. If the child failed one optotype in line 1 logMAR (20/200) to 0.48 logMAR (20/60), two or more optotypes in line logMAR 0.40 (20/50) to 0 logMAR (20/20), or three or more optotypes in line j0.10 logMAR (20/16) to j0.30 logMAR (20/10), the visual acuity was recorded as the line above. The child was required to tell each optotype within 5 seconds; if a longer time was taken, the doctor would proceed to another optotype. The child was not allowed to squeeze or squint during the examination. A table-mounted KR-8800 (Topcon, Tokyo, Japan), a commonly used autorefractor in Asia, with a measurement range of j20 to +20 diopters (D), was used to obtain autorefraction data from each eye beginning with the right eye, before and after cycloplegia. Three consecutive readings of sphere, cylinder, and axis of each eye were taken, and the average was used for analysis. The autorefractometer was calibrated before each examination. Cycloplegia was induced by instillation of 0.5% tropicamide, the most widely used cycloplegic agent for myopia screening in China. One drop was administered every 5 minutes for a total of 5 drops. Although tropicamide inhibits accommodation less effectively than cyclopentolate,31 many studies have reported it to be an effective cycloplegic agent for myopic children in dark
Optometry and Vision Science, Vol. 90, No. 12, December 2013
Copyright © American Academy of Optometry. Unauthorized reproduction of this article is prohibited.
Myopia Screening by Acuity and Noncycloplegic AutorefractionVMa et al.
irises.32Y34 Cycloplegia was considered complete when the pupil dilated to 6 mm or greater and a light reflex was absent. Autorefraction measurements were carried out at an average of 40 minutes after the first drop of tropicamide, if adequate cycloplegia was obtained. If a child did not reach full dilation at the time of measurement, one or two additional drops were given to get full dilation. Because unaided visual acuity, NCAR, and cycloplegic autorefraction results were closely related between the right and the left eyes (Spearman correlations were 0.745, 0.817, and 0.845 for UCVA, NCAR, and CAR, respectively; values of p G 0.001), the right eye was chosen for data analysis. In this study, myopia was defined as a spherical equivalent refraction (sphere + 0.5 cylinder, SER) of at least j0.5 D. Cycloplegic autorefraction was used as the reference standard because this method of examination has been proven to be reliable and stable for detecting refractive error in young children.18,19 Receiver-operating characteristic (ROC) curves were constructed to obtain the best cutoff value, area under the curve (AUC), and Youden index for each test (UCVA and NCAR). Sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and the 95% confidence intervals (95% CIs) were also calculated for every referral threshold (MedCalc 11.4.2.0). A specificity of 90% was set for the purpose of screening,27,35 and the criteria with the highest sensitivity were selected as optimal. We then combined UCVA and NCAR in serial order (children were referred if they failed both tests) to explore the optimal criteria and their corresponding sensitivity, specificity, and Youden Index for myopia screening. Children were divided into three age groups as follows: 6 to 7 years, 8 to 10 years, and 11 to 12 years to further explore the effectiveness of this combination. A paired t test was used to compare the difference in SER before and after cycloplegia.
1481
corresponding sensitivities, specificities, predictive values, and Youden Indexes are listed in Table 1. Fig. 2 shows the ROC curve for the NCAR test. The AUC was 0.940 (95% CI, 0.928 to 0.951), larger than the AUC for the UCVA test (p G 0.0001), indicating a greater ability of the NCAR test to discriminate myopia compared with the UCVA. The best cutoff point was an SER of j0.75 D, with a sensitivity of 88.55% and a specificity of 86.13% (Table 2). This criterion can be accepted when sensitivity and specificity are regarded as equally important. To reduce the cost of unnecessary referral after screening, a higher specificity was preferred in this study because the screening was applied to a population on a large scale.36 In this study, we selected a specificity of 90% as the standard for the screening tests, which is commonly used for mass screening.27,35 Using this standard, the criterion for SER of j1.0 D or less had a sensitivity of 78.50% and a specificity of 91.91%. Performing two diagnostic tests in serial order can theoretically improve specificity but can reduce sensitivity. Namely, the specificity of the combined test is higher than that of either test alone, but the sensitivity of the combined test is lower than that of either test.37 Combining UCVA with NCAR has been reported to have increased the specificity without significantly dropping the sensitivity in preschool children.29,30 Therefore, to achieve more satisfactory sensitivity results along with the required specificity, we tried different combinations of UCVA with NCAR tests in serial order. The best combination was UCVA less than or equal to 0 logMAR (20/20) and NCAR less than or equal to j0.75 D, which had a sensitivity of 84.35% and a specificity of 90.50%. Compared with either using UCVA less than or equal to 0 logMAR (20/20) or NCAR less than or equal to j0.75 D as referral criteria, the combined test increased specificity without
RESULTS Of the 1639 eyes included in the analysis, 428 (26.11%) eyes were diagnosed as being myopic by cycloplegic autorefraction. The median UCVA was 0 logMAR (20/20) for the study eyes (N = 1639). There were 256 children aged 6 to 7 years, 931 children aged 8 to 10 years, and 452 children aged 11 to 12 years. Using the data from the right eye, the prevalence rate of myopia (SER less than or equal to j0.5 D) was 8.98% for ages 6 to 7 years, 22.23% for ages 8 to 10 years, and 43.81% for ages 11 to 12 years. The mean SER before cycloplegia was j0.51 T 1.20 D for all children and 0.00 T1.31 D after cycloplegia. The mean SER was an average of 0.51 D more myopic when measured with NCAR than with cycloplegia autorefraction (p G 0.001). Fig. 1 shows the ROC curve for UCVA. The area under the ROC curve was 0.851 (95% CI, 0.833 to 0.868). The cutoff point was a visual acuity of 0.20 logMAR (20/32), with the corresponding sensitivity of 63.55% and specificity of 93.97%. With a UCVA of 0.10 logMAR (20/25), the most commonly used criterion for vision screening in China, the sensitivity is 80.14% and the specificity is 76.71%. The relatively low specificity means that the UCVA test is likely to incorrectly refer 23.29% of children with normal vision. Other referral criteria and their
FIGURE 1. Receiver operating characteristic curve of UCVA for myopia screening in schoolchildren. The points for UCVA of 0.2 logMAR (Snellen 20/32) and 0.1 logMAR (Snellen 20/25) were labeled. LogMAR, logarithm of the minimum angle of resolution; UCVA, uncorrected visual acuity.
Optometry and Vision Science, Vol. 90, No. 12, December 2013
Copyright © American Academy of Optometry. Unauthorized reproduction of this article is prohibited.
1482 Myopia Screening by Acuity and Noncycloplegic AutorefractionVMa et al. TABLE 1.
Sensitivity, specificity, PPV, NPV, and Youden Index of UCVA for schoolchildren (N = 1639) LogMAR (Snellen) 0.40 (20/50) 0.30 (20/40) 0.20 (20/32) 0.10 (20/25) 0 (20/20) j0.10 (20/16)
Sensitivity
95% CI
Specificity
95% CI
PPV
95% CI
NPV
95% CI
Youden Index
30.84 47.43 63.55 80.14 92.29 98.83
26.5Y35.5 42.6Y52.3 58.8Y68.1 76.0Y83.8 89.3Y94.6 97.3Y99.6
99.01 97.52 93.97 76.71 38.23 6.94
98.3Y99.5 96.5Y98.3 92.5Y95.2 74.2Y79.1 35.5Y41.0 5.6Y8.5
91.7 87.1 78.8 54.9 34.6 27.3
85.9Y95.6 82.1Y91.1 74.1Y83.0 50.9Y58.8 31.8Y37.4 25.1Y29.6
80.2 84.0 87.9 91.6 93.3 94.4
78.1Y82.2 82.0Y85.9 86.0Y89.7 89.7Y93.3 90.8Y95.4 87.4Y98.2
0.2985 0.4495 0.5752 0.5685 0.3052 0.0577
LogMAR, logarithm of the minimum angle of resolution; NPV, negative predictive value; PPV, positive predictive value; UCVA, uncorrected visual acuity.
substantially decreasing sensitivity. In addition, at a specificity of 90%, the sensitivities are 63.55%, 78.50%, and 84.35%, respectively for UCVA, NCAR, and combination of the two (Table 3). Moreover, the Youden Index, which describes the accuracy of a test, was higher for the combination test than for either the UCVA test or the NCAR test alone, suggesting that the combination test is better than the individual tests for diagnosing myopia. We then conducted further analyses in the three different age groups (Table 3). When the specificity was set to 90%, sensitivity was always the lowest for the UCVA test alone and the highest for the combination of the UCVA and NCAR tests in all age groups. The Youden Index of the combination test was also the highest for all age groups, indicating that the accuracy of the combination test is better than that of the UCVA or NCAR test alone. The best criteria for the combination test were a little different among the three age groups. Although applying the combination of UCVA less than or equal to 0 logMAR (20/20) and NCAR less than or equal to j0.75 D can achieve a specificity close to or more than 90% in all age groups, the sensitivity was not always the highest. For children aged 6 to 7 years, the combined use of UCVA less than or equal to 0.1 logMAR (20/25) and NCAR less than or equal to j0.5 D obtained a sensitivity of 91.30%, which was higher than that for the combination of UCVA less than or equal to 0 logMAR (20/20) and NCAR less than or equal to j0.75 D. This result suggests that, to obtain the optimal effect, different criteria need to be established for different age groups when applying the combination test for myopia screening.
the UCVA may be better than that of the Chinese Standard Logarithmic Visual Acuity E Chart likely because of the use of letter-by-letter rather than line-by-line scoring.13 Although the Topcon KR-8800 has been commonly used for refractive measurements in Asia,38,39 the accuracy of this autorefractor has not been reported. In this study, at the optimal cutoff point of j0.75 D (SER less than or equal to j0.75 D), the NCAR test using the Topcon KR-8800 had a sensitivity of 88.55% and a specificity of 86.13%. These results are similar to those of the NCAR test that used the Topcon KR-8100 when screening for significant refractive error in Taiwanese preschool children.30 Compared with other types of autorefractors, the NCAR by the Topcon KR-8800 had a relatively high sensitivity and specificity when used for myopia screening in school-aged children, albeit not the highest.40,41 Moreover, our study shows that the accuracy of the NCAR test is better than that of the UCVA test for detecting myopia, which is demonstrated by the larger AUC of the former. This phenomenon has also been
DISCUSSION In our study, the optimal criterion for UCVA using the Chinese Standard Logarithmic Visual Acuity E Chart was 0.2 logMAR (20/32), with a sensitivity of 63.55% and a specificity of 93.97%. Compared with the visual acuity of 0.1 logMAR (20/25), the most frequently used criterion for vision screening in China, it has a satisfactory specificity, however, a lower sensitivity. The prevalence of myopia is high among schoolchildren in China, and a test with a sensitivity of 63.55% would not detect 36.45% of myopic children. The accuracy of the UCVA test in the study seems to be lower than that reported in other studies.12,15,16 This discrepancy might result from the different standards for defining myopia and different methods to record visual acuity. For instance, sometimes an SER of j1.0 D or higher was used to define myopia.16 The accuracy of the EDTRS chart for measuring
FIGURE 2. Receiver operating characteristic curve of NCAR for myopia screening in schoolchildren. The points for SER of j0.5 D, j0.75 D, j1.0 D, and j1.25 D were labeled. NCAR, noncycloplegic autorefraction; SER, spherical equivalent refraction.
Optometry and Vision Science, Vol. 90, No. 12, December 2013
Copyright © American Academy of Optometry. Unauthorized reproduction of this article is prohibited.
Myopia Screening by Acuity and Noncycloplegic AutorefractionVMa et al.
1483
TABLE 2.
Sensitivity, specificity, PPV, NPV, their 95% CI, and Youden Index of NCAR for schoolchildren (N = 1639) SER, D
Sensitivity
95% CI
Specificity
95% CI
PPV
95% CI
NPV
95% CI
Youden Index
j1.5 j1.25 j1 j0.75 j0.5 j0.25
52.10 66.36 78.50 88.55 95.56 98.83
47.3Y56.9 61.7Y70.8 74.3Y82.3 85.1Y91.4 93.2Y97.3 97.3Y99.6
97.69 95.87 91.91 86.13 75.64 57.64
96.7Y98.5 94.6Y96.9 90.2Y93.4 84.1Y88.0 73.1Y78.0 54.8Y60.4
88.8 85.0 77.4 69.3 58.1 45.2
84.3Y92.5 80.7Y88.7 73.2Y81.3 65.2Y73.1 54.4Y61.8 42.0Y48.4
85.2 89.0 92.4 95.5 98.0 99.3
83.3Y87.1 87.1Y90.6 90.7Y93.8 94.1Y96.7 96.8Y98.8 98.3Y99.8
0.4979 0.6223 0.7041 0.7468 0.7120 0.5647
NCAR, noncycloplegic autorefraction; NPV, negative predictive value; PPV, positive predictive value; SER, spherical equivalent refraction.
observed in studies that focused on screening for refractive error in preschool children.29,30 Attempts to measure refractive status without cycloplegia can lead to significant errors, such as an overestimation of myopia because of the strong accommodative reserve, especially in children younger than 12 years.26 In our study, NCAR overestimates the extent of myopia by, on average, approximately 0.50 D. This disadvantage of NCAR hampers its widespread application in myopia screening in school-aged children. Whether satisfactory screening efficacy can be obtained through the combined use of UCVA and NCAR for myopia screening in schoolchildren has not yet been reported in previous research. Theoretically, with a high sensitivity but a low specificity, using the criterion of UCVA less than or equal to 0 logMAR (20/20) alone cannot discriminate hyperopic and astigmatic eyes whose UCVA were 0 logMAR (20/20) or less. However, combining UCVA with NCAR may eliminate these false-positive conditions
to some extent. Meanwhile, using a single NCAR test can overestimate myopic refraction in school-aged children, mistakenly diagnosing as myopic those children who are emmetropic or hyperopic. However, those people usually have a UCVA of 0 logMAR (20/20) or more, which can be distinguished by combining NCAR with the UCVA test. Combining the two methods in serial order increased the specificity without extensively decreasing the sensitivity in schoolaged children; these results are similar to those found in previous studies in preschool children.29,30 In addition, at a specificity of 90%, the sensitivity for the combination test was always the highest among the three tests in all age groups. Hence, the combination test should not only reduce the cost of unnecessary referrals to ophthalmic departments but also detect more myopic children than either of the two tests alone. In the present study of 1639 subjects, compared with the UCVA test of 0.8 (20/25), the most widely used method for myopia screening in China,21
TABLE 3.
Sensitivity, specificity, and criteria for UCVA, NCAR, and combined use of UCVA and NCAR when specificity is set to 90% (N = 1639) Sensitivity, %
Specificity, %
Youden Index
Criteria
All (N = 1639) UCVA NCAR Combination*
63.55 78.50 84.35
93.97 91.91 90.50
0.575 0.704 0.749
0.2 logMAR (Snellen 20/32) SER e j1.0 D UCVA e 0 logMAR (20/20) and SER e j0.75 D
6Y7 y (n = 256) UCVA NCAR Combination 1* Combination 2*
69.49 86.96 86.96 91.30
94.15 91.42 93.99 92.37
0.636 0.784 0.810 0.837
0.2 logMAR (20/32) SER e j0.75 D UCVA e 0 logMAR (20/20) and SER e j0.75 D UCVA e 0.1 logMAR (20/25) and SER e j0.5 D
8Y10 y (n = 931) UCVA NCAR Combination 1* Combination 2*
67.15 77.78 80.68 85.02
93.78 91.71 93.37 89.23
0.609 0.695 0.741 0.743
0.2 logMAR (20/32) SER e j1.0 D UCVA e 0.1 logMAR (20/25) and SER e j0.5 D UCVA e 0 logMAR (20/20) and SER e j0.75 D
11Y12 y (n = 452) UCVA NCAR Combination*
59.09 78.79 83.33
94.49 90.94 90.94
0.536 0.697 0.743
0.2 logMAR (20/32) SER e j1.0 D UCVA e 0 logMAR (20/20) and SER e j0.75 D
*Combination means combined use of UCVA and NCAR in serial order (children are referred if they failed both tests). LogMAR, logarithm of the minimum angle of resolution; NCAR, noncycloplegic autorefraction; SER, spherical equivalent refraction; UCVA, uncorrected visual acuity. Optometry and Vision Science, Vol. 90, No. 12, December 2013
Copyright © American Academy of Optometry. Unauthorized reproduction of this article is prohibited.
1484 Myopia Screening by Acuity and Noncycloplegic AutorefractionVMa et al.
the combination test resulted in 18 more true-positive subjects (true-positive number: UCVA, n = 343; combination test, n = 361), and 167 fewer false-positive subjects with myopia (falsepositive number: UCVA, n = 282; combination test, n = 155). The cost for an autorefractor is approximately 80,000 RMB (12,878 US dollars) and the cost for an unnecessary referral to the ophthalmology department is estimated to be 100 RMB (16 US dollars) per child, according to the average outpatient service charge in the ophthalmic department of Shanghai. Therefore, with 7851 subjects included, the cost for the autorefractor could be compensated, rendering the combination test to be a cost-effective measurement. Moreover, without cycloplegia, the procedure is simple and time saving, and does not require additional personnel to administer the cycloplegic agents. Like most studies, there are some limitations in our study. First, the combination test cannot discriminate children with pseudomyopia who might have UCVA less than or equal to 0 logMAR (20/20) and NCAR less than or equal to j0.75 D but SER of less than j0.5 D after cycloplegia. Second, some children with true myopia might have NCAR greater than j0.75 D and UCVA greater than 0 logMAR (20/20) but are left undetected by the combination criterion. Moreover, as sensitivity and specificity results change with prevalence rates of diseases,42 the results of the combined test might not be effectively applied in populations with distinctly different prevalence rates of myopia. Last, but not least, the best criterion for screening myopia by the combination test cannot be arbitrarily determined in the present study. As shown in the results, the optimal criteria for screening myopia were not consistent among the different age groups. Different criteria may be established according to different age groups. In conclusion, the combined use of the UCVA and NCAR in serial order achieved higher sensitivity than either of the two tests alone, when the specificity was set at 90%. This new method can be applied to large-scale myopia screening in Chinese schoolchildren aged 6 to 12 years.
ACKNOWLEDGMENTS This research is supported by Funding of Scientific Research for young researchers by Shanghai Health Bureau (No. 2010Y113). No commercial interest or potential conflict of interest exists for any author. The first two authors contributed equally to this work and are considered co-first authors. Received March 30, 2013; accepted August 30, 2013.
REFERENCES 1. Zhao J, Pan X, Sui R, Munoz SR, Sperduto RD, Ellwein LB. Refractive Error Study in Children: results from Shunyi District, China. Am J Ophthalmol 2000;129:427Y35. 2. Congdon N, Wang Y, Song Y, Choi K, Zhang M, Zhou Z, Xie Z, Li L, Liu X, Sharma A, Wu B, Lam DS. Visual disability, visual function, and myopia among rural chinese secondary school children: the Xichang Pediatric Refractive Error Study (X-PRES)Vreport 1. Invest Ophthalmol Vis Sci 2008;49:2888Y94. 3. He M, Zeng J, Liu Y, Xu J, Pokharel GP, Ellwein LB. Refractive error and visual impairment in urban children in southern China. Invest Ophthalmol Vis Sci 2004;45:793Y9.
4. Pan CW, Ramamurthy D, Saw SM. Worldwide prevalence and risk factors for myopia. Ophthalmic Physiol Opt 2012;32:3Y16. 5. He M, Zheng Y, Xiang F. Prevalence of myopia in urban and rural children in mainland China. Optom Vis Sci 2009;86:40Y4. 6. Edwards MH, Lam CS. The epidemiology of myopia in Hong Kong. Ann Acad Med Singapore 2004;33:34Y8. 7. Saw SM, Carkeet A, Chia KS, Stone RA, Tan DT. Component dependent risk factors for ocular parameters in Singapore Chinese children. Ophthalmology 2002;109:2065Y71. 8. Lin LL, Shih YF, Hsiao CK, Chen CJ. Prevalence of myopia in Taiwanese schoolchildren: 1983 to 2000. Ann Acad Med Singapore 2004;33:27Y33. 9. Mathers M, Keyes M, Wright M. A review of the evidence on the effectiveness of children’s vision screening. Child Care Health Dev 2010;36:756Y80. 10. Morgan IG, Ohno-Matsui K, Saw SM. Myopia. Lancet 2012;379: 1739Y48. 11. Walline JJ, Lindsley K, Vedula SS, Cotter SA, Mutti DO, Twelker JD. Interventions to slow progression of myopia in children. Cochrane Database Syst Rev 2011:CD004916. 12. Tong L, Saw SM, Tan D, Chia KS, Chan WY, Carkeet A, Chua WH, Hong CY. Sensitivity and specificity of visual acuity screening for refractive errors in school children. Optom Vis Sci 2002; 79:650Y7. 13. Tong L, Saw SM, Chan ES, Yap M, Lee HY, Kwang YP, Tan D. Screening for myopia and refractive errors using LogMAR visual acuity by optometrists and a simplified visual acuity chart by nurses. Optom Vis Sci 2004;81:684Y91. 14. Luo HD, Gazzard G, Liang Y, Shankar A, Tan DT, Saw SM. Defining myopia using refractive error and uncorrected logMAR visual acuity 90.3 from 1334 Singapore school children ages 7Y9 years. Br J Ophthalmol 2006;90:362Y6. 15. Leone JF, Mitchell P, Morgan IG, Kifley A, Rose KA. Use of visual acuity to screen for significant refractive errors in adolescents: is it reliable? Arch Ophthalmol 2010;128:894Y9. 16. O’Donoghue L, Rudnicka AR, McClelland JF, Logan NS, Saunders KJ. Visual acuity measures do not reliably detect childhood refractive errorYan epidemiological study. PLoS One 2012;7:e34441. 17. National Standard of People’s Republic of China. Standard Logarithmic Visual Acuity E Chart (GB 11533-1989). Beijing, China: China Standard Press; 1990. 18. Choong YF, Chen AH, Goh PP. A comparison of autorefraction and subjective refraction with and without cycloplegia in primary school children. Am J Ophthalmol 2006;142:68Y74. 19. Harvey EM, Miller JM, Dobson V, Tyszko R, Davis AL. Measurement of refractive error in Native American preschoolers: validity and reproducibility of autorefraction. Optom Vis Sci 2000; 77:140Y9. 20. Pedamallu SB, Reddy KS, Pedamallu R, Pedamallu CS. Reliability of cycloplegic autorefractor measurements to determine spherical and astigmatic refractive errors in young children. Internet J Ophthalmol Visual Sci 2008;6(1):doi: 10.5580/1663. 21. Fotouhi A, Hashemi H, Khabazkhoob M, Mohammad K. The prevalence of refractive errors among schoolchildren in Dezful, Iran. Br J Ophthalmol 2007;91:287Y92. 22. Williams C, Miller L, Northstone K, Sparrow JM. The use of noncycloplegic autorefraction data in general studies of children’s development. Br J Ophthalmol 2008;92:723Y4. 23. Fotouhi A, Morgan IG, Iribarren R, Khabazkhoob M, Hashemi H. Validity of noncycloplegic refraction in the assessment of refractive errors: the Tehran Eye Study. Acta Ophthalmol 2012;90:380Y6.
Optometry and Vision Science, Vol. 90, No. 12, December 2013
Copyright © American Academy of Optometry. Unauthorized reproduction of this article is prohibited.
Myopia Screening by Acuity and Noncycloplegic AutorefractionVMa et al. 24. Harvey EM, Miller JM, Dobson V, Tyszko R, Davis AL. Measurement of refractive error in Native American preschoolers: validity and reproducibility of autorefraction. Optom Vis Sci 2000; 77:140Y9. 25. Zhao J, Mao J, Luo R, Li F, Pokharel GP, Ellwein LB. Accuracy of noncycloplegic autorefraction in school-age children in China. Optom Vis Sci 2004;81:49Y55. 26. Fotedar R, Rochtchina E, Morgan I, Wang JJ, Mitchell P, Rose KA. Necessity of cycloplegia for assessing refractive error in 12-year-old children: a population-based study. Am J Ophthalmol 2007;144:307Y9. 27. Schmidt P, Maguire M, Dobson V, Quinn G, Ciner E, Cyert L, Kulp MT, Moore B, Orel-Bixler D, Redford M, Ying GS. Comparison of preschool vision screening tests as administered by licensed eye care professionals in the Vision In Preschoolers Study. Vision in Preschoolers (VIP) Study Group. Ophthalmology 2004;111:637Y50. 28. Jorge J, Queiros A, Gonzalez-Meijome J, Fernandes P, Almeida JB, Parafita MA. The influence of cycloplegia in objective refraction. Ophthalmic Physiol Opt 2005;25:340Y5. 29. Miller JM, Harvey EM, Dobson V. Visual acuity screening versus noncycloplegic autorefraction screening for astigmatism in Native American preschool children. J AAPOS 1999;3:160Y5. 30. Lai YH, Tseng HY, Hsu HT, Chang SJ, Wang HZ. Uncorrected visual acuity and noncycloplegic autorefraction predict significant refractive errors in Taiwanese preschool children. Ophthalmology 2013;120:271Y6. 31. Egashira SM, Kish LL, Twelker JD, Mutti DO, Zadnik K, Adams AJ. Comparison of cyclopentolate versus tropicamide cycloplegia in children. Optom Vis Sci 1993;70:1019Y26. 32. Hamasaki I, Hasebe S, Kimura S, Miyata M, Ohtsuki H. Cycloplegic effect of 0.5% tropicamide and 0.5% phenylephrine mixed eye drops: objective assessment in Japanese schoolchildren with myopia. Jpn J Ophthalmol 2007;51:111Y5. 33. Fan DS, Rao SK, Ng JS, Yu CB, Lam DS. Comparative study on the safety and efficacy of different cycloplegic agents in children with darkly pigmented irides. Clin Experiment Ophthalmol 2004; 32:462Y7.
1485
34. Chen X, Lin Z, Lai X, Fang B, Li C, Li L. An objective investigation on the cycloplegic effects of tropicamide. China J Optom Ophthalmol 2008;10:135Y8. 35. Vision in Preschoolers (VIP) Study Group. Preschool vision screening tests administered by nurse screeners compared with lay screeners in the vision in preschoolers study. Invest Ophthalmol Vis Sci 2005; 46:2639Y48. 36. Wilson JM, Jungner G. Principles and Practice of Screening for Disease. World Health Organization Public Health Paper No. 34. Geneva, Switzerland: World Health Organization; 1968. 37. Weinstein S, Obuchowski NA, Lieber ML. Clinical evaluation of diagnostic tests. AJR Am J Roentgenol 2005;184:14Y9. 38. Fotouhi A, Hashemi H, Yekta AA, Mohammad K, Khoob MK. Characteristics of astigmatism in a population of schoolchildren, Dezful, Iran. Optom Vis Sci 2011;88:1054Y9. 39. Shen P, Zheng Y, Ding X, Liu B, Congdon N, Morgan I, He M. Biometric measurements in highly myopic eyes. J Cataract Refract Surg 2013;39:180Y7. 40. 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;9:59Y70. 41. Rogers DL, Neely DE, Chapman JB, Plager DA, Sprunger DT, Sondhi N, Roberts GJ, Ofner S. Comparison of the MTI Photoscreener and the Welch-Allyn SureSight autorefractor in a tertiary care center. J AAPOS 2008;12:77Y82. 42. Brenner H, Gefeller O. Variation of sensitivity, specificity, likelihood ratios and predictive values with disease prevalence. Stat Med 1997; 16:981Y91.
Jianfeng Zhu No. 380 Kangding Rd, Jingan District Shanghai Eye Disease Prevention and Treatment Center Shanghai 200040 China e-mail: [email protected]
Optometry and Vision Science, Vol. 90, No. 12, December 2013
Copyright © American Academy of Optometry. Unauthorized reproduction of this article is prohibited.
