- Email: [email protected]
The Sonksen logMAR test of visual acuity: II. Age norms from 2 years 9 months to 8 years
The Sonksen logMAR test of visual acuity: II. Age norms from 2 years 9 months to 8 years Patricia M. Sonksen, MB BS, MD, FRCPCH, FRCP,a,b Angie M. Wade, MSc, PhD,a Ruth Proffitt, DBO, BSc,c Sally Heavens,c and Alison T. Salt, MSc, FRACP, FRCPCHa,b PURPOSE METHODS
RESULTS
CONCLUSIONS
To establish age norms and interocular differences in visual acuity between 2 years 9 months and 8 years for the Sonksen logMAR Test. Cross-sectional population-based study. Binocular measures of linear visual acuity were achieved in 2,940 children and monocular measures were achieved in 2,820 right eyes and 2,821 left eyes, respectively. Measures for both right and left eyes were achieved by 2,807. Asymmetric logistic models were used to construct smoothly changing age-related centile curves showing how visual acuity changes with age in a normative population sample. All curves demonstrated an increase in visual acuity with age that was steepest between 2 years 9 months and 5 years 3 months. Equivalent centiles for linear visual acuity were better when viewed binocularly than monocularly; the difference was least between the 95th centiles ( best levels) and greatest between the 5th centiles (worst levels). There were no clinically significant differences between group measures of visual acuity from right and left eyes—average within child difference 0.0095 logMAR units, 95% CI, 0.0059-0.013. Interocular differences did not vary significantly with age ( p ⫽ 0.73). The 90th and 95th centiles for interocular difference were 0.125 and 0.175 log units, respectively. This study demonstrates how visual acuity varied with age for the Sonksen logMAR Test and presents the findings in the clinically useful format of centile charts. ( J AAPOS 2008;12:18-22)
T
his article presents the third stage of the development of the Sonksen logMAR Test—the development of age norms for linear visual acuity from 2 years 9 months to 8 years in the form of centile charts derived from a population-based normative sample. The first two stages— design of the test and test protocol together with testability and reliability studies—are presented in our companion article.1 Realization that visual acuity for letter and symbol test displays changes rapidly during early childhood has led clinicians and researchers to advocate the development of age norms for young children.2-4 Monitoring changes in visual acuity from childhood into adulthood would be enhanced by
Author affiliations: aInstitute of Child Health, University College, London, United Kingdom; bGreat Ormond Street Hospital NHS Trust, London, United Kingdom; c Lifespan NHS Trust, Cambridge, United Kingdom This study was jointly conducted at Institute of Child Health, University College, London, and Lifespan NHS Trust, Cambridge, United Kingdom. Novomed, UK, produces the Sonksen logMAR Test commercially. Once production costs are recovered by the company, a research fund held at University College, London by Sonksen and Salt will receive 10% of profits from sales. Supported by The NHS Executive, Anglia and Oxford Division, United Kingdom. Submitted November 12, 2006. Revision accepted April 30, 2007. Published online July 28, 2007. Reprint requests: Dr. Patricia M. Sonksen, The Wolfson Centre, Institute of Child Health, University College, Mecklenburgh Square, London, WC1N 2AP UK (email: [email protected]). Copyright © 2008 by the American Association for Pediatric Ophthalmology and Strabismus. 1091-8531/2008/$35.00 ⫹ 0 doi:10.1016/j.jaapos.2007.04.019
18
test displays that adhere closely to standard recommendations.5 The construct and scoring of the logMAR scale has been acknowledged as superior for measuring visual acuity and is better able to accommodate the rapid changes in visual acuity of young children than the Snellen scale.5-9 LogMAR tests available for preschool children when the Sonksen logMAR Test was designed did not conform fully to standard specifications in one or more respects10-12; hence, we designed the Sonksen logMAR Test to accord as closely as possible to standard ETDRS specifications. Normative datasets represent the general population with respect to the target biological skill and therefore contain individuals with impairments of that skill.13 In the context of visual acuity those above the 90th centile have “exceptionally” good vision with respect to the majority and are not of clinical concern, whereas those below the 10th centile have “exceptionally” poor vision and are the ones most likely to be the concern of ophthalmic services. Some of the early studies of age-related changes in visual acuity used large population samples at specific age bands with variously crowded Snellen-based charts or single letter displays.14-16 More recently our group published proportions achieving 20/20 and 20/30 for the Snellenbased Sonksen Silver Acuity System17 in a populationbased study of 2 years 6 months to 9-year-olds.4,18 Unfortunately, recent studies, using logMAR-based tests, have, with the exception of Robeai et al,19 applied visual or social/developmental sampling criteria that deviate from the definition of a normative dataset (above).11,12,20,21
Journal of AAPOS
Volume 12 Number 1 / February 2008
Sonksen et al
19
Table 1. Distribution in age bands of achievement of each test of visual acuity Number and (%) in age groups Acuity Test
24 to ⬍36 mo
36 to ⬍48 mo
48 to ⬍60 mo
60 to ⬍72 mo
72 to ⬍84 mo
84 to 104 mo
Total
Singles binocular Linear binocular Linear monocular right eye Linear monocular left eye Total
134 (92) 117 (80) 74 (50)
567 (97) 558 (95) 507 (86)
709 (99.7) 708 (99.6) 688 (97)
635 (99.9) 636 (100) 632 (99.3)
554 (99.9) 554 (99.9) 552 (99.4)
367 (100) 367 (100) 367 (100)
2966 (99) 2940 (98) 2820 (94)
74 (50)
505 (86)
689 (97)
633 (99.4)
553 (99.6)
367 (100)
2821 (94)
147 (100)
587 (100)
711 (100)
636 (100)
555 (100)
367 (100)
2991 (100)
Tests of visual acuity are shown in the left hand column in the order in which they were carried out in the protocol. Age bands in years are shown along the top with the number and ( percentage) of childern in each column that achieved each test of visual acuity. Note that the numbers and percentages are derived from Table 1 of the companion paper1; two sets of two 6 month age bands have been truncated into two year bands (2 years to ⬍2 years 6 months with 2 years 6 months to ⬍3 years) and (3 years to ⬍3 years 6 months with 3 years 6 months to ⬍4 years) as the age norms in this paper are derived from actual age in months rather than age bands. The 84-⬍96 month-old-age group is expanded to include those between 8 years and 8 years, 6 months.
Anisometropia and amblyopia characterized by differences in visual acuity between the eyes are particularly important to the long-term visual health of children.22 The degree of interocular difference that should give rise to clinical concern at different ages is therefore a major concern of clinicians and a focus in some recent studies.21,23-25 We therefore examined how interocular differences in visual acuity vary with age.
Materials and Methods The project was collaborative between the Institute of Child Health, London, and Lifespan Health Care Trust, Cambridge, with the approval of their respective ethics committees. Test design, test protocol, recruitment, and population have been described in detail previously.1 The cohort of 2991 children aged 2 to 8 years 6 months was cross-sectional and population based. The developmental features for optimal participation of preschool children that were introduced by Sonksen into the Sonksen Silver Acuity System were included in the test protocol.17 Test displays in ETDRS format in the prototype test booklets of the Sonksen logMAR Test ranged from logMAR 0.775 to – 0.300. Every display had four letters to a line, each with a value of 0.025 logMAR units surrounded by crowding bars. Testing was carried out by five orthoptists according to the standard protocol, which included training in matching letters, equivalent to the binocular pretest of some studies.11,12,20 Prior to occlusion, binocular single and linear visual acuity were measured followed by linear visual acuity of the right then left eyes. Every letter correctly identified contributed to the score. The dataset for binocular visual acuity for crowded single optotypes was based on viewing only one optotype of each size in comparison to those for linear visual acuity—four of each size. The test distance was 3 m (10 feet). Snellen equivalents for logMAR units are cited in the text with the nominator as 20: eg, 0.700 (20/100), 0.600 (20/80), 0.500 (20/64), 0.400 (20/50), 0.300 (20/40), 0.200 (20/32), 0.100 (20/25), 0.000 (20/20), ⫺0.100 (20/16), ⫺0.200 (20/12.5), ⫺0.300 (20/10).
Statistical Methods Centiles were created for visual acuity for single and linear displays viewed binocularly and for linear displays viewed mo-
Journal of AAPOS
nocularly using previously developed methods for ordinal ratings.18 The increased validity obtained via treatment of these outcomes as ordinal as opposed to numeric was previously established.26 Briefly, asymmetric logistic models were used to describe the age-related changes in the logit of the cumulative probabilities at each age with model parameters estimated via maximum likelihood. Full details of the fitting process and alternative methods of presentation of the results have been published.26 Presentation of the results as centile curves is similar to that used for other physiological measurements such as height and weight and hence easily interpretable for the clinician. The age range of the centile graphs—2 years 9 months to 8 years—is narrower than the testability statistics in the companion article,1 as the statistical method for compilation of norms requires data above and below the limiting ages to give reasonable estimation of the centiles at those extremes. The average within-child difference in linear visual acuity between the right and left eyes measured monocularly is presented with 95% confidence intervals. Regression analysis was used to determine any age-trends in the interocular differences.
Results The distribution of gender, ethnicity, and socioeconomic status in the 2,991 children comprising the normative dataset was in line with that of the UK.1 Table 1 gives the distribution according to age for achievement of measures of binocular visual acuity for single letter displays and the three measures of linear visual acuity. Centile curves of linear visual acuity derived from the normative dataset for the test displays of the Sonksen logMAR Test viewed binocularly and monocularly are presented in Figures 1 and 2, respectively, for children aged 2 years 9 months to 8 years. The average within-child difference between eyes was 0.0095 (95% CI, 0.00590.013) in the 2807 children who achieved monocular measures for both right and left eyes. The median difference was zero and there was no evidence of any trend ( p ⫽ 0.73). We therefore present the curves of the left eye only as representative of monocular centiles. Both binocular and monocular curves of linear visual acuity increase most rapidly in the 30 months between 2
20
Sonksen et al
Volume 12 Number 1 / February 2008
FIG 1. Age norms for binocular linear visual acuity from 2 years 9 months to 8 years. Visual acuity is shown on the left-hand vertical axis and age along the horizontal axis. Visual acuity improves from the bottom to the top of the chart; thus, logMAR scores with a negative sign indicate the best levels of visual acuity. The area under the 10th centile is shaded mid-gray on the centile charts to assist clinicians in identifying the 10% of children with the poorest visual acuities for age. Key to centile curve lines: Thickest continuous ⫽ 50th; mid-thickness continuous ⫽ 5th and 95th; thinnest continuous ⫽ 25th and 50th; dashed ⫽ 10th and 90th. The centile curves are also notated on the right-hand vertical axis. Usage: The point of intersection of a horizontal line from the logMAR score with a vertical from the age in years and months pinpoints the position of an individual (or group) measure of visual acuity, relative to that of children of the same age in the general population. For example, the binocular visual acuity of a child scoring logMAR 0.100 (20/25) at age: ● 2 years 11 months would be better than average (75th centile) for age ● 3 years 4 months would be average ( just below 50th centile) for age ● 6 years 3 months would be in the “exceptionally” poor range ( below 5th centile, which falls within the definition of “exceptionally” as below the 10th).
years 9 months and 5 years 3 months. The increase in both measures during this period is almost three times that occurring during the subsequent 33 months to 8 years. The equivalent centiles for binocular and monocular visual acuity (Figures 1 and 2) indicate a trend for better binocular than monocular visual acuity at every age. The 5th, 50th, and 95th centiles appear superimposed in eSupplement 1 (available online at jaapos.org) accompanying this article and show that disparity is least between the 95th centiles— best levels of visual acuity and greatest between the 5th centiles ( poorest levels). The centile curves derived for binocular visual acuity for single optotypes appear in e-Supplement 2 (available online at jaapos.org). Equivalent binocular centiles for single and linear displays (e-Supplement 2 and Figure 1, respectively) indicate better visual acuities for single than for linear displays under the age of 5 years; the difference increases with decreasing age, being maximal (approximately 0.150 log units for the 50th centile) at 2 years 9 months. Interocular differences did not vary significantly with age. The difference in visual acuity of each eye was 0.000 log units in 22%, 0.025 in 26%, 0.050 in 20%, 0.075 in 12%, 0.100 in 10%, and 0.125 or more in 10%. The 90th
centile for interocular difference was 0.125 and the 95th 0.175 logMAR units.
Discussion Population-based age norms show how binocular and monocular linear visual acuity changes with age between 2 years 9 months and 8 years for the Sonksen logMAR Test. The centiles show how population values are distributed at each age. They can be used in the same way as growth charts by clinicians to determine the position of an individual (or group) measure of visual acuity relative to that of children of the same age in the general population, using the method described in the legend to Figure 1. The centiles demonstrate that visual acuity changes most rapidly between 2 years 9 months and 5 years 3 months and highlight the disadvantages of calculating limits of normality using means and standard deviations derived from grouping children into 6- or 12-month age blocks. In the same way as logMAR scaling is more sensitive than Snellen scaling for measuring visual acuity, centile curves reflect the changes that occur within the time span of an age block, which
Journal of AAPOS
Volume 12 Number 1 / February 2008
Sonksen et al
21
FIG 2. Age norms for monocular linear visual acuity from 2 years 9 months to 8 years. Monocular (left eye) linear visual acuity is presented in the same format as in Figure 1.
is particularly important during early childhood when change is rapid. The wide range of centiles from the 5th to the 95th will allow clinicians and researchers to select centiles most pertinent to the nature of their practice or aims of their research. The strengths of the study are a test design1,27 that adheres to international guidelines for standardized logMAR tests,5-9 test administration according to a developmentally strong1,4,17,27 standardized test protocol,8,5 and age norms that are derived from a large and representative normative population sample13 using statistically strong methodology.18,26 Provision of age norms together with test design in close accord with the adult standard complies with recommendations for improving comparability of measurement from childhood into adulthood.3,28,29 The limitations of the study are that the centiles of visual acuity for crowded singles viewed binocularly are based on only one presentation at each level and that there were no serial measurements within the dataset with which to identify the extent to which children track centiles. The latter needs to be kept in mind when using cross-sectional centiles to monitor longitudinal changes. Fern and Manny recommended that tests for children should include displays of letters to logMAR ⫺0.300 (20/ 10).3 Single and linear displays to this level were included in the Sonksen logMAR Test as in some other tests used in recent studies.10,19,25 Recently, monocular visual acuity (ETDRS chart) was found to be significantly better than 0.000 (20/20) in teenagers.25 Our normative curves
Journal of AAPOS
(ETDRS format) reveal that 50% achieve monocular visual acuities of logMAR 0.000 (20/20) or better by 4 years 6 months and that the 10% with best acuities (90th centile and above) achieve ⫺0.100 (20/16) by this age. Similarly, in a previous population-based study, with the Sonksen Silver Acuity System—Snellen scaling, limit 20/20 —50% of children achieved monocular linear visual acuities of 20/20 by the age of 4 years 6 months.4,18 Fiftieth centiles equate to medians and also to means if, and only if, data distribution is symmetric. Our 50th centiles tend to show better visual acuity than the means reported in the literature; however, it is difficult to compare our findings directly with these studies for a variety of reasons. The statistical approach is different and many of the estimates rely on small sample sizes.11,21,23,24 Our measure of crowded single visual acuity was binocular so is not comparable with the monocular HOTV singles findings of the Amblyopia Treatment Study12 and Vision in Preschoolers (2003) Study.11 We have found no studies that provide norms for binocular linear visual acuity. In all studies of monocular linear visual acuity, except that of Robeai and coworkers,19 the samples are not normative; they are skewed intentionally to children without visual defects,21,23,24 or to those with visual defects plus or minus socioeconomic and health problems.11,20 Although the tests used have logMAR scaling and linear format, crowding and optotype formats vary.17,19-21,23,24 The developmental content of test instruction and of the response task affects a child’s confidence in the test
22
Sonksen et al
Volume 12 Number 1 / February 2008
setting and the depth of their understanding of the test task; both will affect concentration and willingness to continue during testing. Only the Sonksen Silver Acuity System4,17 and Sonksen logMAR Test protocols1,27 include full 3-m binocular linear testing before the challenge of occlusion is confronted. Binocular summation is likely to contribute to the trend for binocular visual acuity to be better than monocular visual acuity at every age. Similar differences have been reported in children for grating acuity30 and acuity measured with Cardiff cards.31 Our findings are in agreement with those in the literature in that we found no clinically significant difference in the measurements for right and left eyes19-21,24,25 or in interocular differences with age.21,24,25,32 Our 95th and 90th centiles for interocular difference are similar to those estimated for 3- and 4-year-olds by Shea and Gaccon.21 The ability of each of these centiles to identify the individual clinical conditions that feature unequal vision in childhood remains to be established.
Conclusions We have achieved stage 3 in the development of the Sonksen logMAR Test—provision of age norms for clinical use in the form of centile charts. The 10th centile and area below it identify the 10% of children with the poorest visual acuities for age; further studies are needed to establish the relationship of visual acuities in this range to visual defects considered relevant to ophthalmic practice. Interocular differences did not vary significantly with age.
Acknowledgments We thank the research orthoptists and the staff of participating schools and nurseries for their enthusiastic help and support throughout the period of data collection. References 1. Salt AT, Wade AM, Proffitt R, Heavens S, Sonksen PM. The Sonksen logMAR Test of Visual Acuity: I. Testability and reliability. J AAPOS In press 2007. 2. Simons K. Visual acuity norms in young children. Surv Ophthalmol 1983;28:84-91. 3. Fern KD, Manny RE. Visual acuity in preschool children: A review. Am J Optom Physiol Optics 1986;63:319-45. 4. Salt AT, Sonksen PM, Wade A, Jayatunga R. The maturation of linear acuity and compliance with the Sonksen Silver Acuity System of young children. Dev Med Ch Neurol 1995;37:505-14. 5. Ferris FL, Bailey I. Standardizing the measurement of visual acuity for clinical research studies: Guidelines from the Eye Care Technology Forum. Ophthalmology 1996;103:181-2. 6. Bailey I, Lovie J. New design principles for visual acuity letter charts. Am J Optom Physiol Opt 1976;53:740-5. 7. National Academy of Sciences—National Research Council Committee on Vision. Report of working Group 39: Recommended standard procedures for the clinical measurement and specification of visual acuity. Adv Ophthalmol 1980;41:103-48.
8. Ferris FL, Kassoff A, Bresnick GH, Bailey I. New Visual Acuity charts for clinical research. Am J Ophthalmol 1982;94:91-6. 9. Consilium Ophthalmologicum Universale: Visual Functions Committee: Visual acuity measurement standard. Ital J Ophthalmol 1988;1:15. 10. McGraw PV, Winn B. Glasgow Acuity Cards: A new test for the measurement of letter acuity in children. Ophthal Physiol Opt 1993; 13:400-4. 11. Vision in Preschoolers Study Group. Threshold visual acuity testing of preschool children using the crowded HOTV and Lea symbols Acuity Tests. J AAPOS 2003;7:396-9. 12. Holmes JM, Beck RW, Repka MX, Leske DA, Kraker RT, Blair C, et al. The Amblyopia Treatment Study Visual Acuity Testing Protocol. Arch Ophthalmol 2001;119:1345-53. 13. Anastasi A, Urbina S. Psychological Testing. 7th ed. Upper Saddle River (NJ): Prentice Hall; 1997. Chapters 1 and 2. 14. Alberman EA, Butler NR, Sheridan MD. Visual acuity of a national sample (1958 cohort) at 7 years. Dev Med Child Neurol 1971;13:9-14. 15. Peckham CS. Vision in childhood. Br Med Bull 1986;42:150-4. 16. Pott JWR, van Hoff-van Duin J. The Rotterdam C-Chart: Norm values of visual acuity and intra-ocular differences in 5-year-old children. Behav Brain Res 1992;49:141-7. 17. Sonksen PM, Silver J. The Sonksen Silver Acuity System: Test System and Instruction Manual. Windsor (Berkshire, UK): Keeler Ltd.; 1988. 18. Wade A, Ades AE, Salt AT, Jayatunga R, Sonksen PM. Age-related standards for ordinal data: Modeling the changes in visual acuity from 2 to 9 years of age. Stats Med 1995;14:257-66. 19. Robaei D, Rose K, Ojaimi E, Kifley A, Huynh S, Mitchell P. Visual acuity and the causes of visual loss in a population-based sample of 6-year-old Australian children. Ophthalmology 2005;112:1275-82. 20. Vision in Preschoolers Study Group. Preschool visual acuity screening with HOTV and Lea symbols: Testability and between-test agreement. Optom Vision Sci 2004;81:678-83. 21. Shea, SJ, Gaccon, L. In the absence of strabismus what constitutes a visual deficit in children? Br J Ophthalmol 2006;90:40-3. 22. Hartmann E, Dobson V, Hainlain L, Marsh-Tootle W, Quinn GE, Ruttum MS, et al. Preschool vision screening: Summary of a task force report. Pediatrics 2000;106:1105-12. 23. Simmers AJ, Gray LS, Spowart K. Screening for amblyopia: A comparison of paediatric letter tests. Br J Ophthalmol 1997;81: 465-9. 24. Stewart C. Comparison of Snellen and log-based acuity scores for school-aged children. Br Orthopt J 2000;57:32-8. 25. Ohlsson J, Villarreal G. Normal visual acuity in 17-18 year olds. Acta Ophthalmol Scand 2005;83:487-91. 26. Wade AM, Salt AT, Proffitt R, Heavens SJ, Sonksen PM. Likelihood-based modeling of age related normal ranges for ordinal measurements: Changes in visual acuity through early childhood. Stats Med 2004;23:3623-40. 27. Sonksen PM. Manual for the Sonksen logMAR Test of Visual Acuity. Novomed, Maidstone, Kent, UK; 2006. 28. Atkinson J, Pimm-Smith E, Evans C, Harding G, Braddick OJ. Visual crowding in young children. Doc Ophthalmol Proc Series 1986;45:210-3. 29. Morad Y, Werker E, Nemet P. Visual acuity tests using chart, line, and single optotype in healthy and amblyopic children. J AAPOS 1999;3:94-7. 30. Vital Durand F, Hullo A. 500 visual acuity tests in infants with Teller acuity cards. Ophtalmologie 1990;4:208-11. 31. Adoh TO, Woodhouse JM. The Cardiff acuity test for measuring visual acuity development in toddlers. Vis Res 1994;34:555-60. 32. Brown B, Yap MK. Differences in visual acuity between the eyes: Determination of normal limits in a clinical population. Ophthal Physiol Opt 1995;15:163-9.
Journal of AAPOS
RESULTS
CONCLUSIONS
To establish age norms and interocular differences in visual acuity between 2 years 9 months and 8 years for the Sonksen logMAR Test. Cross-sectional population-based study. Binocular measures of linear visual acuity were achieved in 2,940 children and monocular measures were achieved in 2,820 right eyes and 2,821 left eyes, respectively. Measures for both right and left eyes were achieved by 2,807. Asymmetric logistic models were used to construct smoothly changing age-related centile curves showing how visual acuity changes with age in a normative population sample. All curves demonstrated an increase in visual acuity with age that was steepest between 2 years 9 months and 5 years 3 months. Equivalent centiles for linear visual acuity were better when viewed binocularly than monocularly; the difference was least between the 95th centiles ( best levels) and greatest between the 5th centiles (worst levels). There were no clinically significant differences between group measures of visual acuity from right and left eyes—average within child difference 0.0095 logMAR units, 95% CI, 0.0059-0.013. Interocular differences did not vary significantly with age ( p ⫽ 0.73). The 90th and 95th centiles for interocular difference were 0.125 and 0.175 log units, respectively. This study demonstrates how visual acuity varied with age for the Sonksen logMAR Test and presents the findings in the clinically useful format of centile charts. ( J AAPOS 2008;12:18-22)
T
his article presents the third stage of the development of the Sonksen logMAR Test—the development of age norms for linear visual acuity from 2 years 9 months to 8 years in the form of centile charts derived from a population-based normative sample. The first two stages— design of the test and test protocol together with testability and reliability studies—are presented in our companion article.1 Realization that visual acuity for letter and symbol test displays changes rapidly during early childhood has led clinicians and researchers to advocate the development of age norms for young children.2-4 Monitoring changes in visual acuity from childhood into adulthood would be enhanced by
Author affiliations: aInstitute of Child Health, University College, London, United Kingdom; bGreat Ormond Street Hospital NHS Trust, London, United Kingdom; c Lifespan NHS Trust, Cambridge, United Kingdom This study was jointly conducted at Institute of Child Health, University College, London, and Lifespan NHS Trust, Cambridge, United Kingdom. Novomed, UK, produces the Sonksen logMAR Test commercially. Once production costs are recovered by the company, a research fund held at University College, London by Sonksen and Salt will receive 10% of profits from sales. Supported by The NHS Executive, Anglia and Oxford Division, United Kingdom. Submitted November 12, 2006. Revision accepted April 30, 2007. Published online July 28, 2007. Reprint requests: Dr. Patricia M. Sonksen, The Wolfson Centre, Institute of Child Health, University College, Mecklenburgh Square, London, WC1N 2AP UK (email: [email protected]). Copyright © 2008 by the American Association for Pediatric Ophthalmology and Strabismus. 1091-8531/2008/$35.00 ⫹ 0 doi:10.1016/j.jaapos.2007.04.019
18
test displays that adhere closely to standard recommendations.5 The construct and scoring of the logMAR scale has been acknowledged as superior for measuring visual acuity and is better able to accommodate the rapid changes in visual acuity of young children than the Snellen scale.5-9 LogMAR tests available for preschool children when the Sonksen logMAR Test was designed did not conform fully to standard specifications in one or more respects10-12; hence, we designed the Sonksen logMAR Test to accord as closely as possible to standard ETDRS specifications. Normative datasets represent the general population with respect to the target biological skill and therefore contain individuals with impairments of that skill.13 In the context of visual acuity those above the 90th centile have “exceptionally” good vision with respect to the majority and are not of clinical concern, whereas those below the 10th centile have “exceptionally” poor vision and are the ones most likely to be the concern of ophthalmic services. Some of the early studies of age-related changes in visual acuity used large population samples at specific age bands with variously crowded Snellen-based charts or single letter displays.14-16 More recently our group published proportions achieving 20/20 and 20/30 for the Snellenbased Sonksen Silver Acuity System17 in a populationbased study of 2 years 6 months to 9-year-olds.4,18 Unfortunately, recent studies, using logMAR-based tests, have, with the exception of Robeai et al,19 applied visual or social/developmental sampling criteria that deviate from the definition of a normative dataset (above).11,12,20,21
Journal of AAPOS
Volume 12 Number 1 / February 2008
Sonksen et al
19
Table 1. Distribution in age bands of achievement of each test of visual acuity Number and (%) in age groups Acuity Test
24 to ⬍36 mo
36 to ⬍48 mo
48 to ⬍60 mo
60 to ⬍72 mo
72 to ⬍84 mo
84 to 104 mo
Total
Singles binocular Linear binocular Linear monocular right eye Linear monocular left eye Total
134 (92) 117 (80) 74 (50)
567 (97) 558 (95) 507 (86)
709 (99.7) 708 (99.6) 688 (97)
635 (99.9) 636 (100) 632 (99.3)
554 (99.9) 554 (99.9) 552 (99.4)
367 (100) 367 (100) 367 (100)
2966 (99) 2940 (98) 2820 (94)
74 (50)
505 (86)
689 (97)
633 (99.4)
553 (99.6)
367 (100)
2821 (94)
147 (100)
587 (100)
711 (100)
636 (100)
555 (100)
367 (100)
2991 (100)
Tests of visual acuity are shown in the left hand column in the order in which they were carried out in the protocol. Age bands in years are shown along the top with the number and ( percentage) of childern in each column that achieved each test of visual acuity. Note that the numbers and percentages are derived from Table 1 of the companion paper1; two sets of two 6 month age bands have been truncated into two year bands (2 years to ⬍2 years 6 months with 2 years 6 months to ⬍3 years) and (3 years to ⬍3 years 6 months with 3 years 6 months to ⬍4 years) as the age norms in this paper are derived from actual age in months rather than age bands. The 84-⬍96 month-old-age group is expanded to include those between 8 years and 8 years, 6 months.
Anisometropia and amblyopia characterized by differences in visual acuity between the eyes are particularly important to the long-term visual health of children.22 The degree of interocular difference that should give rise to clinical concern at different ages is therefore a major concern of clinicians and a focus in some recent studies.21,23-25 We therefore examined how interocular differences in visual acuity vary with age.
Materials and Methods The project was collaborative between the Institute of Child Health, London, and Lifespan Health Care Trust, Cambridge, with the approval of their respective ethics committees. Test design, test protocol, recruitment, and population have been described in detail previously.1 The cohort of 2991 children aged 2 to 8 years 6 months was cross-sectional and population based. The developmental features for optimal participation of preschool children that were introduced by Sonksen into the Sonksen Silver Acuity System were included in the test protocol.17 Test displays in ETDRS format in the prototype test booklets of the Sonksen logMAR Test ranged from logMAR 0.775 to – 0.300. Every display had four letters to a line, each with a value of 0.025 logMAR units surrounded by crowding bars. Testing was carried out by five orthoptists according to the standard protocol, which included training in matching letters, equivalent to the binocular pretest of some studies.11,12,20 Prior to occlusion, binocular single and linear visual acuity were measured followed by linear visual acuity of the right then left eyes. Every letter correctly identified contributed to the score. The dataset for binocular visual acuity for crowded single optotypes was based on viewing only one optotype of each size in comparison to those for linear visual acuity—four of each size. The test distance was 3 m (10 feet). Snellen equivalents for logMAR units are cited in the text with the nominator as 20: eg, 0.700 (20/100), 0.600 (20/80), 0.500 (20/64), 0.400 (20/50), 0.300 (20/40), 0.200 (20/32), 0.100 (20/25), 0.000 (20/20), ⫺0.100 (20/16), ⫺0.200 (20/12.5), ⫺0.300 (20/10).
Statistical Methods Centiles were created for visual acuity for single and linear displays viewed binocularly and for linear displays viewed mo-
Journal of AAPOS
nocularly using previously developed methods for ordinal ratings.18 The increased validity obtained via treatment of these outcomes as ordinal as opposed to numeric was previously established.26 Briefly, asymmetric logistic models were used to describe the age-related changes in the logit of the cumulative probabilities at each age with model parameters estimated via maximum likelihood. Full details of the fitting process and alternative methods of presentation of the results have been published.26 Presentation of the results as centile curves is similar to that used for other physiological measurements such as height and weight and hence easily interpretable for the clinician. The age range of the centile graphs—2 years 9 months to 8 years—is narrower than the testability statistics in the companion article,1 as the statistical method for compilation of norms requires data above and below the limiting ages to give reasonable estimation of the centiles at those extremes. The average within-child difference in linear visual acuity between the right and left eyes measured monocularly is presented with 95% confidence intervals. Regression analysis was used to determine any age-trends in the interocular differences.
Results The distribution of gender, ethnicity, and socioeconomic status in the 2,991 children comprising the normative dataset was in line with that of the UK.1 Table 1 gives the distribution according to age for achievement of measures of binocular visual acuity for single letter displays and the three measures of linear visual acuity. Centile curves of linear visual acuity derived from the normative dataset for the test displays of the Sonksen logMAR Test viewed binocularly and monocularly are presented in Figures 1 and 2, respectively, for children aged 2 years 9 months to 8 years. The average within-child difference between eyes was 0.0095 (95% CI, 0.00590.013) in the 2807 children who achieved monocular measures for both right and left eyes. The median difference was zero and there was no evidence of any trend ( p ⫽ 0.73). We therefore present the curves of the left eye only as representative of monocular centiles. Both binocular and monocular curves of linear visual acuity increase most rapidly in the 30 months between 2
20
Sonksen et al
Volume 12 Number 1 / February 2008
FIG 1. Age norms for binocular linear visual acuity from 2 years 9 months to 8 years. Visual acuity is shown on the left-hand vertical axis and age along the horizontal axis. Visual acuity improves from the bottom to the top of the chart; thus, logMAR scores with a negative sign indicate the best levels of visual acuity. The area under the 10th centile is shaded mid-gray on the centile charts to assist clinicians in identifying the 10% of children with the poorest visual acuities for age. Key to centile curve lines: Thickest continuous ⫽ 50th; mid-thickness continuous ⫽ 5th and 95th; thinnest continuous ⫽ 25th and 50th; dashed ⫽ 10th and 90th. The centile curves are also notated on the right-hand vertical axis. Usage: The point of intersection of a horizontal line from the logMAR score with a vertical from the age in years and months pinpoints the position of an individual (or group) measure of visual acuity, relative to that of children of the same age in the general population. For example, the binocular visual acuity of a child scoring logMAR 0.100 (20/25) at age: ● 2 years 11 months would be better than average (75th centile) for age ● 3 years 4 months would be average ( just below 50th centile) for age ● 6 years 3 months would be in the “exceptionally” poor range ( below 5th centile, which falls within the definition of “exceptionally” as below the 10th).
years 9 months and 5 years 3 months. The increase in both measures during this period is almost three times that occurring during the subsequent 33 months to 8 years. The equivalent centiles for binocular and monocular visual acuity (Figures 1 and 2) indicate a trend for better binocular than monocular visual acuity at every age. The 5th, 50th, and 95th centiles appear superimposed in eSupplement 1 (available online at jaapos.org) accompanying this article and show that disparity is least between the 95th centiles— best levels of visual acuity and greatest between the 5th centiles ( poorest levels). The centile curves derived for binocular visual acuity for single optotypes appear in e-Supplement 2 (available online at jaapos.org). Equivalent binocular centiles for single and linear displays (e-Supplement 2 and Figure 1, respectively) indicate better visual acuities for single than for linear displays under the age of 5 years; the difference increases with decreasing age, being maximal (approximately 0.150 log units for the 50th centile) at 2 years 9 months. Interocular differences did not vary significantly with age. The difference in visual acuity of each eye was 0.000 log units in 22%, 0.025 in 26%, 0.050 in 20%, 0.075 in 12%, 0.100 in 10%, and 0.125 or more in 10%. The 90th
centile for interocular difference was 0.125 and the 95th 0.175 logMAR units.
Discussion Population-based age norms show how binocular and monocular linear visual acuity changes with age between 2 years 9 months and 8 years for the Sonksen logMAR Test. The centiles show how population values are distributed at each age. They can be used in the same way as growth charts by clinicians to determine the position of an individual (or group) measure of visual acuity relative to that of children of the same age in the general population, using the method described in the legend to Figure 1. The centiles demonstrate that visual acuity changes most rapidly between 2 years 9 months and 5 years 3 months and highlight the disadvantages of calculating limits of normality using means and standard deviations derived from grouping children into 6- or 12-month age blocks. In the same way as logMAR scaling is more sensitive than Snellen scaling for measuring visual acuity, centile curves reflect the changes that occur within the time span of an age block, which
Journal of AAPOS
Volume 12 Number 1 / February 2008
Sonksen et al
21
FIG 2. Age norms for monocular linear visual acuity from 2 years 9 months to 8 years. Monocular (left eye) linear visual acuity is presented in the same format as in Figure 1.
is particularly important during early childhood when change is rapid. The wide range of centiles from the 5th to the 95th will allow clinicians and researchers to select centiles most pertinent to the nature of their practice or aims of their research. The strengths of the study are a test design1,27 that adheres to international guidelines for standardized logMAR tests,5-9 test administration according to a developmentally strong1,4,17,27 standardized test protocol,8,5 and age norms that are derived from a large and representative normative population sample13 using statistically strong methodology.18,26 Provision of age norms together with test design in close accord with the adult standard complies with recommendations for improving comparability of measurement from childhood into adulthood.3,28,29 The limitations of the study are that the centiles of visual acuity for crowded singles viewed binocularly are based on only one presentation at each level and that there were no serial measurements within the dataset with which to identify the extent to which children track centiles. The latter needs to be kept in mind when using cross-sectional centiles to monitor longitudinal changes. Fern and Manny recommended that tests for children should include displays of letters to logMAR ⫺0.300 (20/ 10).3 Single and linear displays to this level were included in the Sonksen logMAR Test as in some other tests used in recent studies.10,19,25 Recently, monocular visual acuity (ETDRS chart) was found to be significantly better than 0.000 (20/20) in teenagers.25 Our normative curves
Journal of AAPOS
(ETDRS format) reveal that 50% achieve monocular visual acuities of logMAR 0.000 (20/20) or better by 4 years 6 months and that the 10% with best acuities (90th centile and above) achieve ⫺0.100 (20/16) by this age. Similarly, in a previous population-based study, with the Sonksen Silver Acuity System—Snellen scaling, limit 20/20 —50% of children achieved monocular linear visual acuities of 20/20 by the age of 4 years 6 months.4,18 Fiftieth centiles equate to medians and also to means if, and only if, data distribution is symmetric. Our 50th centiles tend to show better visual acuity than the means reported in the literature; however, it is difficult to compare our findings directly with these studies for a variety of reasons. The statistical approach is different and many of the estimates rely on small sample sizes.11,21,23,24 Our measure of crowded single visual acuity was binocular so is not comparable with the monocular HOTV singles findings of the Amblyopia Treatment Study12 and Vision in Preschoolers (2003) Study.11 We have found no studies that provide norms for binocular linear visual acuity. In all studies of monocular linear visual acuity, except that of Robeai and coworkers,19 the samples are not normative; they are skewed intentionally to children without visual defects,21,23,24 or to those with visual defects plus or minus socioeconomic and health problems.11,20 Although the tests used have logMAR scaling and linear format, crowding and optotype formats vary.17,19-21,23,24 The developmental content of test instruction and of the response task affects a child’s confidence in the test
22
Sonksen et al
Volume 12 Number 1 / February 2008
setting and the depth of their understanding of the test task; both will affect concentration and willingness to continue during testing. Only the Sonksen Silver Acuity System4,17 and Sonksen logMAR Test protocols1,27 include full 3-m binocular linear testing before the challenge of occlusion is confronted. Binocular summation is likely to contribute to the trend for binocular visual acuity to be better than monocular visual acuity at every age. Similar differences have been reported in children for grating acuity30 and acuity measured with Cardiff cards.31 Our findings are in agreement with those in the literature in that we found no clinically significant difference in the measurements for right and left eyes19-21,24,25 or in interocular differences with age.21,24,25,32 Our 95th and 90th centiles for interocular difference are similar to those estimated for 3- and 4-year-olds by Shea and Gaccon.21 The ability of each of these centiles to identify the individual clinical conditions that feature unequal vision in childhood remains to be established.
Conclusions We have achieved stage 3 in the development of the Sonksen logMAR Test—provision of age norms for clinical use in the form of centile charts. The 10th centile and area below it identify the 10% of children with the poorest visual acuities for age; further studies are needed to establish the relationship of visual acuities in this range to visual defects considered relevant to ophthalmic practice. Interocular differences did not vary significantly with age.
Acknowledgments We thank the research orthoptists and the staff of participating schools and nurseries for their enthusiastic help and support throughout the period of data collection. References 1. Salt AT, Wade AM, Proffitt R, Heavens S, Sonksen PM. The Sonksen logMAR Test of Visual Acuity: I. Testability and reliability. J AAPOS In press 2007. 2. Simons K. Visual acuity norms in young children. Surv Ophthalmol 1983;28:84-91. 3. Fern KD, Manny RE. Visual acuity in preschool children: A review. Am J Optom Physiol Optics 1986;63:319-45. 4. Salt AT, Sonksen PM, Wade A, Jayatunga R. The maturation of linear acuity and compliance with the Sonksen Silver Acuity System of young children. Dev Med Ch Neurol 1995;37:505-14. 5. Ferris FL, Bailey I. Standardizing the measurement of visual acuity for clinical research studies: Guidelines from the Eye Care Technology Forum. Ophthalmology 1996;103:181-2. 6. Bailey I, Lovie J. New design principles for visual acuity letter charts. Am J Optom Physiol Opt 1976;53:740-5. 7. National Academy of Sciences—National Research Council Committee on Vision. Report of working Group 39: Recommended standard procedures for the clinical measurement and specification of visual acuity. Adv Ophthalmol 1980;41:103-48.
8. Ferris FL, Kassoff A, Bresnick GH, Bailey I. New Visual Acuity charts for clinical research. Am J Ophthalmol 1982;94:91-6. 9. Consilium Ophthalmologicum Universale: Visual Functions Committee: Visual acuity measurement standard. Ital J Ophthalmol 1988;1:15. 10. McGraw PV, Winn B. Glasgow Acuity Cards: A new test for the measurement of letter acuity in children. Ophthal Physiol Opt 1993; 13:400-4. 11. Vision in Preschoolers Study Group. Threshold visual acuity testing of preschool children using the crowded HOTV and Lea symbols Acuity Tests. J AAPOS 2003;7:396-9. 12. Holmes JM, Beck RW, Repka MX, Leske DA, Kraker RT, Blair C, et al. The Amblyopia Treatment Study Visual Acuity Testing Protocol. Arch Ophthalmol 2001;119:1345-53. 13. Anastasi A, Urbina S. Psychological Testing. 7th ed. Upper Saddle River (NJ): Prentice Hall; 1997. Chapters 1 and 2. 14. Alberman EA, Butler NR, Sheridan MD. Visual acuity of a national sample (1958 cohort) at 7 years. Dev Med Child Neurol 1971;13:9-14. 15. Peckham CS. Vision in childhood. Br Med Bull 1986;42:150-4. 16. Pott JWR, van Hoff-van Duin J. The Rotterdam C-Chart: Norm values of visual acuity and intra-ocular differences in 5-year-old children. Behav Brain Res 1992;49:141-7. 17. Sonksen PM, Silver J. The Sonksen Silver Acuity System: Test System and Instruction Manual. Windsor (Berkshire, UK): Keeler Ltd.; 1988. 18. Wade A, Ades AE, Salt AT, Jayatunga R, Sonksen PM. Age-related standards for ordinal data: Modeling the changes in visual acuity from 2 to 9 years of age. Stats Med 1995;14:257-66. 19. Robaei D, Rose K, Ojaimi E, Kifley A, Huynh S, Mitchell P. Visual acuity and the causes of visual loss in a population-based sample of 6-year-old Australian children. Ophthalmology 2005;112:1275-82. 20. Vision in Preschoolers Study Group. Preschool visual acuity screening with HOTV and Lea symbols: Testability and between-test agreement. Optom Vision Sci 2004;81:678-83. 21. Shea, SJ, Gaccon, L. In the absence of strabismus what constitutes a visual deficit in children? Br J Ophthalmol 2006;90:40-3. 22. Hartmann E, Dobson V, Hainlain L, Marsh-Tootle W, Quinn GE, Ruttum MS, et al. Preschool vision screening: Summary of a task force report. Pediatrics 2000;106:1105-12. 23. Simmers AJ, Gray LS, Spowart K. Screening for amblyopia: A comparison of paediatric letter tests. Br J Ophthalmol 1997;81: 465-9. 24. Stewart C. Comparison of Snellen and log-based acuity scores for school-aged children. Br Orthopt J 2000;57:32-8. 25. Ohlsson J, Villarreal G. Normal visual acuity in 17-18 year olds. Acta Ophthalmol Scand 2005;83:487-91. 26. Wade AM, Salt AT, Proffitt R, Heavens SJ, Sonksen PM. Likelihood-based modeling of age related normal ranges for ordinal measurements: Changes in visual acuity through early childhood. Stats Med 2004;23:3623-40. 27. Sonksen PM. Manual for the Sonksen logMAR Test of Visual Acuity. Novomed, Maidstone, Kent, UK; 2006. 28. Atkinson J, Pimm-Smith E, Evans C, Harding G, Braddick OJ. Visual crowding in young children. Doc Ophthalmol Proc Series 1986;45:210-3. 29. Morad Y, Werker E, Nemet P. Visual acuity tests using chart, line, and single optotype in healthy and amblyopic children. J AAPOS 1999;3:94-7. 30. Vital Durand F, Hullo A. 500 visual acuity tests in infants with Teller acuity cards. Ophtalmologie 1990;4:208-11. 31. Adoh TO, Woodhouse JM. The Cardiff acuity test for measuring visual acuity development in toddlers. Vis Res 1994;34:555-60. 32. Brown B, Yap MK. Differences in visual acuity between the eyes: Determination of normal limits in a clinical population. Ophthal Physiol Opt 1995;15:163-9.
Journal of AAPOS