Repeatability of ETDRS Visual Acuity in Children

Abstract

purpose. To evaluate repeatability of the best corrected log minimum angle of resolution (MAR) Early-Treatment Diabetic Retinopathy Study (ETDRS) acuity in a group of 6- to 11-year old children with myopia.

methods. Best corrected monocular visual acuity (VA) of a subset of children (n = 86) enrolled in the Correction of Myopia Evaluation Trial (COMET; mean spherical equivalent refractive error −2.35 D with no more than 1.25 D astigmatism) was measured at baseline and 1 month later with ETDRS logMAR charts. Children started with logMAR 0.4 (6/15 or 20/50) and read each letter on all subsequent lines until they missed all letters in 1 line.

results. At baseline, the mean best corrected logMAR VA was 0.003 ± 0.076 (6/6 or 20/20 ± 3.8 letters) in the right eye and 0.008 ± 0.059 (6/6 or 20/20 ± 2.95 letters) in the left eye. The signed difference between VA measured at baseline and that measured at 1 month was not significantly different from zero in either eye. Repeatability was not associated with age, but a small, statistically significant association with gender was detected in the left eye, with boys approximately 2 letters more variable than girls. The κ statistic (agreement within 1 line) was good to excellent.

conclusions. Based on the 95% limits of agreement, the criterion for a statistically significant change in VA is no more than ±0.15 logMAR (or ±8 letters). This value is similar to those reported for adults and indicates that logMAR VA provides a repeatable measure of acuity in children.

Visual acuity (VA) is used clinically to assess the integrity of the central visual pathway, to monitor the treatment or progression of various ocular conditions (e.g., amblyopia, cataract, macular degeneration), and to guide decisions concerning correction of refractive error. To make many of these decisions, the clinician is required to determine whether the VA of a patient has changed substantially over time between measurements. For a change in acuity to be considered statistically significant, the shift in acuity must exceed the normal expected variability inherent in the measurement. 

Over the past several decades, efforts have been made to reduce the variability and enhance the precision of this fundamental measurement of visual function through improved standardization of acuity charts and testing protocols. Many of these advances were driven by the need for a clinically relevant and reliable outcome measure of visual function to be used in clinical trials. Perhaps the most widely reported and accepted modification has been the ETDRS logMAR chart, which is based on the early work of Bailey and Lovie. ¹ In addition to improvements in chart design, ² the ETDRS protocol ³ standardizes the administration and scoring of VA. With these advances, VA has been used as an outcome measure in clinical trials designed to evaluate a variety of ocular treatments in adults involving the cornea, ^{4 5 6} lens, ⁷ retina, ^{8 9 10} and optic nerve. ^{11 12}  

Despite these advances, there has been surprisingly little energy devoted to applying these improvements in acuity assessment to children and determining the reliability of this key index of visual function when ETDRS charts and procedures are used. The introduction of the Glasgow acuity cards by McGraw et al. ^{13 14} represents one attempt to apply the ETDRS principles to young, primarily preschool children. Although many of the ETDRS principles were incorporated into the design of the Glasgow acuity cards, the spacing of the letters on the Glasgow cards (Glasgow: 0.5-letter diameter spacing between letters; ETDRS: 1-letter diameter spacing between letters), the use of a crowding bar surrounding the 4-letter line array on the Glasgow cards (rather than 5 letters per line without a crowding bar on the ETDRS), the use of 6 letters with the Glasgow acuity cards rather than the 10 possible letters used with the EDTRS chart, and the single line presentation of the Glasgow cards rather than the entire ETDRS chart may result in differences in these two acuity measures in children. Although comparisons between Bailey-Lovie charts and the Glasgow acuity cards have been made in adults with good agreement between the two (95% confidence limits of ±0.07 logMAR, approximately 3 letters), ¹⁴ comparisons between ETDRS charts and Glasgow cards in children have been limited by the younger ages of children who are tested by the Glasgow cards. In two other recent comparisons, only a limited number of children were tested ¹⁵ or Landolt C’s were used instead of Snellen letters (Camparini M, Cassinari P, Ambrosoli D, Porta R, Macaluso C, ARVO Abstract 2068, 2001). Kheterpal et al. ¹⁵ tested the unaided monocular VA of 18 children (mean age, 4.8 years; range, 3.5–8 years) referred for failing a vision screening. The same examiner tested these children on two occasions, separated by 1 week, using a logMAR chart and a matching response. Each letter was scored, and the 95% confidence limit in the right eye was ±0.21 logMAR (approximately 11 letters) and in the left eye was ±0.25 (approximately 13 letters). Campraini et al. measured VA in 40 children 3 to 12 years of age with Landolt C’s in an ETDRS-style chart, using both a standard and modified (ETDRS-Fast) ETDRS protocol (Camparini M, Cassinari P, Ambrosoli D, Porta R, Macaluso C, ARVO Abstract 2068, 2001). They found a standard deviation of the test–retest difference for the standard protocol to be 5.57 letters. The 95% limits of agreement calculated from the standard deviation they report would be 11 letters. This value is the same as that found in the right eye by Kheterpal et al. ¹⁵ The 95% limits of agreement improved to 7 letters with the more rapid modified protocol. 

In this report, the repeatability of best corrected, monocular, logMAR ETDRS acuity was determined in 86 children. These children constitute a subgroup of the 469 children with myopia (age, 6–11 years) participating in COMET. The mean signed difference of the two acuity measures is evaluated by age, gender, examiner, and acuity level. Based on the 95% limits of agreement, the criterion for a statistically significant (and clinically meaningful) change in vision in children 6 to 11 years of age is approximately ±0.15 logMAR (±8 letters, or less than 2 lines) on the logMAR ETDRS chart. 

Methods

Subjects

The 86 children (aged 6–10 years) who participated in this study were among the first of the 469 children enrolled in the Correction of Myopia Evaluation Trial (COMET). COMET is a multicenter clinical trial designed to compare the progression of myopia in children randomized to wear Varilux Comfort (Essilor International., St. Petersburg, FL) progressive addition lenses (add power of +2.00 D) with that in children randomized to single-vision lenses. Additional details regarding COMET design are described by Hyman et al. ¹⁶ One of the eligibility criteria significant in the context of this study is that only children with best corrected logMAR VA of 0.2 (6/9 or 20/32) or better were allowed to enroll in COMET. 

To determine whether an abnormal eye alignment was induced by the random lens assignment, the study’s Data Safety and Monitoring Committee requested that the first children enrolled in COMET be examined 1 month after enrollment. Children did not undergo refraction testing at the 1 month visit but the full 1-month examination included logMAR VA with the child’s habitually worn correction (most often, this was the subjective refraction determined at baseline) and tests of eye alignment. One hundred fifty-two children were examined 1 month after enrollment and were potential subjects for this study of the repeatability of logMAR VA. However, 42 of these children were scheduled for an abbreviated examination that did not include logMAR acuity and were therefore not eligible for this analysis. Twenty-four additional children were excluded because their habitually worn correction differed slightly from the subjective refraction used to measure logMAR acuity at the baseline examination. Thus, the 86 children who had logMAR VA measured both at baseline and 1 month later with identical refractive corrections are the basis of this repeatability evaluation. The characteristics at baseline of these 86 children are shown in Table 1 along with all the 469 children enrolled in COMET. There were no statistically significant differences in the age, gender, refractive error or ethnicity distributions of the 86 children in this study when compared with all COMET children. However, the logMAR VA at baseline of the 86 children reported in this study was significantly different (OD, P = 0.04; OS, P = 0.02) from that of the 469 children enrolled in COMET, with the children in COMET showing slightly poorer VA than those in this study. Although the difference is statistically significant, the mean difference in logMAR VA in the two groups is not clinically meaningful, because it is only 0.018 logMAR in the right eye and 0.02 in the left eye (1 letter or less). Additional details regarding the baseline characteristics of the children enrolled in COMET are reported by Gwiazda et al. ¹⁷  

COMET and its protocols conform to the tenets of the Declaration of Helsinki. Before enrollment, children and parents were informed of the nature of the trial, and informed consent was obtained from the parent or guardian after the procedures were approved by each local institutional review board. Children also assented to participate in COMET. 

Procedure

Monocular distance VA was measured at baseline with the subjective refraction and 1 month later with the identical refractive correction. The refractive correction was placed in a refractor (model RT1200; Nidek, Gamagori, Japan) with the child appropriately aligned behind the refractor. All examiners were optometrists specifically trained in the modified ETDRS protocol used in COMET. A back-illuminated stand, 4 m from the patient, held the logMAR ETDRS acuity charts (Precision Vision charts 2121 and 2122). Chart lighting ranged from 158 to 279 cd/m² at the four centers and was in compliance with the recommendations for standardizing the measurement of VA (80–320 cd/m²), as summarized by Ferris and Bailey. ^{18 19} Contrast of the letter charts at the four centers averaged 96.9% ± 0.83% with a range of 95.8% to 97.9%. Visual acuity was measured first in the right eye (chart 2121) followed by the left eye (chart 2122), both at baseline and at the second measurement 1 month later. Each of the two charts used ranged from 6/600 (20/200) to 6/3 (20/10) in 0.1 logMAR steps with 5 letters on each line. With this particular chart configuration, each letter was equivalent to 0.02 logMAR. Because all children were free of disease and were tested with their best correction, the ETDRS protocol was modified slightly by having the child begin with the 6/15 (20/50) line and continue until all letters were missed on 1 line, consistent with the recommendations of Carkeet. ²⁰ If the child reached the 6/3 (20/10) without missing an entire line, the 6/3 (20/10) line was the last line attempted. If a letter was missed on the 6/15 (20/50) line, children were directed to start with the 6/30 (20/100) line. Children were required to identify each letter and to guess if not sure. The examiner pointed to each new line and to each letter after the child had incorrectly identified a letter. Visual acuity was determined by scoring 0.02 logMAR for each letter correctly identified. 

To evaluate the repeatability of the best corrected logMAR VA measured in these 86 children with myopia, the VA measured at baseline was subtracted from the VA measured at 1 month. In each child, the difference in VA obtained at these two visits was plotted against the average VA at baseline and 1 month. The SD of the signed difference between repeated measures was used to calculate the 95% limits of agreement. Results are presented in logMAR and number of letters. The number of letters is provided for a clinical reference and applies to the particular chart configuration used in the study. When the number of letters was rounded, it was to the next higher integer, because the number of letters was the unit of measurement. 

Data Analysis

The statistical methods used in the data analysis were based on a hierarchical scheme that comprises data summaries, univariate analyses, and multivariate and modeling analyses. Data summaries were based on estimating distribution parameters such as the mean, median, quartiles, range, and SD for continuous measurements and the frequency and percent for categorical variables. 

Where normality assumptions were met ²¹ statistical comparisons for univariate analyses were based primarily on one or two independent sample(s) t-distributions. For example, a one-sample t-distribution was used to determine whether the difference in VA measurements between the baseline and 1-month visit was significantly different from zero. A two-sample t-test was used to detect a statistically significant difference in the measurements between 1-month and baseline, stratified by whether the measurements were performed on both occasions by the same examiner. Other univariate analyses used the Pearson linear correlation method. Nonparametric methods (Wilcoxon rank sum test) were used in cases in which the normality assumption was not met. 

Multivariate and modeling analyses were based on linear multiple regression techniques to determine the magnitude and the significance of the effect of several covariates (e.g., age, same or different examiner, and gender) on repeatability of VA measurements. To evaluate whether repeatability varied by subgroups, two-factor interaction terms were also included in the multiple linear regression (i.e., different combinations of ages, gender, and same or different examiners). Except for situations involving multiple comparisons, the significance levels for testing a statistically significant difference were preset at α = 0.05. The Dunnett/Bonferroni adjustments were applied when multiple comparisons were made. 

In addition to continuous data analysis, categorical κ ²² was also used for measuring agreement in VA measurements between baseline and 1 month. A κ greater than 0.75 indicates excellent agreement, and a κ between 0.40 and 0.75 suggests the agreement is fair to good. ²³  

Results

At baseline, the mean best corrected logMAR VA was 0.003 ± 0.076 (6/6 or 20/20 ± 3.8 letters) in the right eye and 0.008 ± 0.059 (6/6 or 20/20 ± 2.95 letters) in the left eye. These VAs were not significantly different from each other. At the 1-month visit, the mean best corrected logMAR VA was similar to that obtained at baseline: 0.017 ± 0.083 (6/6 or 20/20 ± 4.2 letters) in the right eye and 0.013 ± 0.074 (6/6 or 20/20 ± 3.7 letters) in the left eye, with no significant difference between the eyes. The signed difference in the best corrected logMAR VA measured at baseline and that measured at the 1-month visit was calculated for each eye and is shown in Figure 1 . The top panel displays data from the right eyes and the bottom panel shows the results in the left eyes. The solid line represents the mean difference: 0.014 ± 0.075 in the right eye and 0.004 ± 0.071 in the left eye. Different symbols are used to mark the positions where more than one child’s acuity is located, with different symbols representing the different numbers of children (see legend to Fig. 1 ). There was no significant difference in the signed difference between the eyes. 

Examination of the signed differences in Figure 1 shows the points to be symmetrically distributed about the mean. In the right eye, the baseline VA was better than the VA at the 1-month visit (positive difference) in 45 children, whereas in 32 children the baseline VA was poorer than the VA at the 1-month visit (negative difference). In the nine remaining children, there was no difference in VA between the 1-month and baseline visits in the right eye. In the left eye, the baseline VA was better than the VA at the 1-month visit in 39 children, whereas in 37 children the baseline VA was poorer than the VA at the 1-month visit. In the 10 remaining children, there was no difference in VA between the 1-month and baseline visits in the left eye. This pattern indicates that the variability in logMAR VA from repeated measures on the same children by multiple examiners is random. The mean signed difference is slightly positive (OD 0.014 ± 0.075; OS 0.004 ± 0.071), but this small difference in VA (less than 1 letter) is not significantly different from zero in either eye (t-test). Because the signed differences are normally distributed, the area within the two dotted lines (mean difference ± 1.96 × SD of the differences) denotes the 95% limits of agreement between the logMAR VA at baseline and the logMAR VA at the 1-month visit. In other words, 95% of the time, the VA measured at two separate times should differ by no more than ±0.15 logMAR (approximately ±8 letters OD, ±7 letters OS) or less than 2 lines. Similar results are obtained by calculating the coefficient of repeatability ²⁴ using the within-subject variance (±0.147 logMAR, or 8 letters OD; ±0.139 logMAR, or 7 letters OS). 

To examine the effects of age, gender, and examiner on the repeatability of VA, a multivariate analysis was performed. Age was investigated as a categorical variable (age 6–9 years versus 10–11 years, inclusive, based on the median age). Looking at the signed difference in VA (1-month minus baseline), there was no significant association with any of the three variables in the right eye. In the left eye, gender was significant (P = 0.032). VA in boys was more variable than girls by 0.037 logMAR (less than 2 letters). The effect of examiner approached significance in the left eye (P = 0.07) with the same examiners showing less variability in test results by 0.03 logMAR (less than 2 letters). The age of the child was not associated with the difference in VA on repeated measures in either eye. 

Another approach to evaluation of the reproducibility of VA measured at two different times is to determine the level of agreement with the κ statistic. Table 2 displays the κ statistic and the 95% confidence limits for agreement between the baseline and 1-month visits to within 1 line of acuity, in each eye of all subjects and by gender and age. In both eyes, agreement to within 1 line is excellent in all subjects, in girls, and in children less than 10 years of age. Agreement in the right eye is good in those older than 10 years (0.71) and in boys (0.68), whereas the left eye shows excellent agreement in these two categories. 

Discussion

The literature is surprisingly sparse on acuity norms in children measured with best correction using ETDRS charts and protocols. Previous reports of corrected acuity in young adults (18–30 years of age) and preschool children (3–5 years) measured using ETDRS principles are summarized in Table 3 . The VAs reported for the young adults are similar and are better than those reported by Sprague et al. ²⁵ for preschool children. The results in Table 3 and those reported in this study are consistent with the notion of Elliott et al. ²⁶ that logMAR acuity improves up to age 29. Thus, the best corrected monocular VA of these 86 children with myopia measured during the baseline examination provide an important reference. Although these logMAR acuities are statistically different from those of the 469 COMET children, they are within 1 letter of the mean of all the children enrolled in COMET. However, caution must be used when applying the acuity reported in the present study to other populations of children, because children with best corrected logMAR VAs worse than 0.2 (6/9 [20/32]) were excluded from enrollment in COMET. 

When the acuity at baseline was compared with that obtained 1 month later with the identical spectacle correction, the mean signed difference was not significantly different from zero for either eye. Given the use of a standard protocol implemented by experienced examiners, it is not surprising that, on average, no difference in acuity was found and indicates that whatever the source of the variability, the variability is random. This lack of bias demonstrates the integrity of the modified ETDRS protocol as used with children in COMET. The result also suggests that a single previous exposure to the test 1 month earlier was not sufficient to improve the measured acuity. The 95% limits of agreement of ±0.015 logMAR (8 letters, or just under 2 lines), provide a useful guideline for the variability expected using this modified ETDRS protocol in children with normal vision tested with either the same or different examiners. However, the variability would be expected to be greater without the uniform application of the standard protocol. 

Several potential sources of the variability indexed by the 95% limits of agreement were analyzed further. There was no statistical difference in the signed difference when the same examiner performed both measures of acuity compared with when different examiners took the two measures. In addition, when the effect of the examiner on the signed difference was explored with a multivariate analysis, the examiner was not associated with the variability in either eye. This result is important in the context of this and other clinical trials involving children, in which multiple examiners are common, and suggests that the standard protocol removes this potential source of bias. It should be noted that the power to detect such a small difference in repeatability with the same examiner compared with different examiners is compromised by the small number of children (59 tested with the same examiner and 27 with a different examiner). However, the power to detect a more clinically meaningful difference of 3 letters when the same examiner obtains both measures of acuity compared with different examiners obtaining the two estimates of acuity is 93% in this sample of children. 

In addition to the examiner, the age and gender of the subjects are other potential sources of variability. Given the limited age range of the subjects in this study (6–11 years), it is not surprising that age was not associated with the sign difference in either the right or left eye. However, caution is advised when generalizing to children younger than age 6 who have more difficulty with chart formats and are frequently tested with isolated letters. ²⁷ A significant association was found for gender. The signed difference in boys showed approximately 2 letters greater variability than that in girls, but only in the left eye. The charts used to measure the acuity differed between the eyes. In addition, VA was always measured in the left eye after that in the right eye. Perhaps the greater variability demonstrated by boys in the left eye is related to the fact that boys tend to fidget more than girls. ^{28 29} Fidgeting has been inversely associated with cognitive ability test scores, and these test scores have been directly associated with concentration in a sample of children (mean age, 9.4 years). ²⁹ However, this potential explanation is only partially consistent with the analysis of agreement (κ statistic). Although the agreement was good ²³ in the right eye of males (κ = 0.68), it was the poorest level of agreement of the repeated measures of acuity. 

With the exception of the good agreement also found in the right eye in children 10 years of age or older (κ = 0.71), all other κ statistics were excellent (κ > 0.75). It is important to recall that in calculating the κ statistic, we used 1 line or 5 letters as the level of agreement to compare the VA measured at baseline with that measured 1 month later. This level of agreement was selected to provide the clinician with an indication of the expected repeatability based on a line of letters rather than individual letters and the agreement is, in general, excellent. 

The criterion for a statistically significant change reported for this subset of COMET children is similar to the data summarized in Table 4 from different populations with a variety of conditions. However, caution must be exercised when generalizing to other populations a criterion derived from children with myopia tested with their best correction. This caution is still warranted, despite the fact that a multivariate analysis that controlled for age and gender indicated that baseline myopia did not influence the magnitude of the difference between the two VA measures (OD P = 0.53; OS: P = 0.11). Children with hyperopia, children with larger amounts of astigmatism, or children tested without correction may not show the same degree of repeatability because of additional sources of error from variable accommodation, spectacle lens aberrations, or blur interpretation. Several investigators have also suggested that estimates of error derived from clinically healthy individuals may not be the same as those for individuals with anomalies. ^{30 31} Therefore, in cases of amblyopia, in which it is important to identify a significant change in acuity in response to treatment, alternative approaches may be needed, particularly if magnitude of difference scores are correlated with raw scores. One approach would be to derive separate estimates of reliability of different subgroups. ³⁰ Another approach would be to determine each individual’s criterion through repeated testing of that individual. ^{32 33 34} Such repeated testing would improve the variability by the square root of the number of estimates. 

This study demonstrates that standard ETDRS charts and a standard protocol can be used in children aged 6 to 11 years to provide a repeatable measure of best corrected monocular VA and that the criterion for a significant change in VA is approximately ±0.15 (±8 letters, or slightly less than 2 lines), similar to that reported in adults. 

Appendix 1

COMET Study Group

STUDY CHAIR’S OFFICE:

New England College of Optometry, Boston, Massachusetts: Jane Gwiazda (Study Chair/Principal Investigator), Kenneth Grice (Study Coordinator), Rosanna Pacella (Research Assistant), and Thomas T. Norton (Consultant, University of Alabama at Birmingham). 

COORDINATING CENTER: Department of Preventive Medicine, University Hospital and Medical Center, Stony Brook, New York: Leslie Hyman (Principal Investigator), M. Cristina Leske (Co-principal Investigator), Mohamed Hussein (Co-Investigator/Biostatistician), Elinor Schoenfeld (Epidemiologist), Lynette Dias (Study Coordinator), Rachel Harrison (Study Coordinator), Elissa Schnall (Assistant Study Coordinator), Allison Schmertz (Project Assistant), Wen Zhu (Lead Programmer), Ahmed Yassin (Analysis), Ying Wang (Analysis), Lauretta Passanant (Project Assistant), and Phyllis Neuschwender (Administrative Assistant). 

CLINICAL CENTERS:

New England College of Optometry, Boston, Massachusetts: Daniel Kurtz (Principal Investigator), Bruce Moore (Optometrist), Robert Owens (Primary Optician), Sheila Martin (Clinic Coordinator), and Stacey Hamlett (Backup Optician). Pennsylvania College of Optometry, Philadelphia, Pennsylvania: Mitchell Scheiman (Principal Investigator), Kathleen Zinzer (Optometrist), Timothy Lancaster (Primary Optician), Theresa Elliot (Backup Optician), and Mariel Torres (Clinic Coordinator). University of Alabama at Birmingham Medical Center, School of Optometry, Birmingham, Alabama: Wendy Marsh-Tootle (Principal Investigator), Bradley S. Bessant (Optometrist), James Raley (Optician), Angela Rawden (Backup Optician), Nicholas Harris (Clinic Coordinator), Cheryl Jackson (Clinic Coordinator), and Trana Mars (Backup Clinic Coordinator). University of Houston College of Optometry, Houston, Texas: Ruth E. Manny (Principal Investigator), Connie Crossnoe (Optometrist), Sheila Deatherage (Optician), Charles Dudonis (Optician), and Sally Henry (Clinic Coordinator). 

NATIONAL EYE INSTITUTE, Bethesda, Maryland:

Donald Everett (Project Director, Collaborative Trials Branch). 

COMMITTEES:

Data Safety and Monitoring: Robert J. Hardy (Chair), Argye Hillis, Don Mutti, Richard Stone, and Sr. Carol Taylor. Executive Committee: Jane Gwiazda (Chair), Leslie Hyman, Wendy Marsh-Tootle Donald Everett. Steering Committee: Jane Gwiazda (Chair), Mohamed Hussein, Leslie Hyman, Daniel Kurtz, M.Christina Leske, Ruth E. Manny, Wendy Mash-Tootle, Mitchell Scheiman, Donald Everett, Thomas T. Norton. 

⁵  COMET Group members are listed in the ^Appendix.

 Presented in part at the American Academy of Optometry, Orlando, Florida, December 2000.

 Supported by National Eye Institute Grants EY11740, EY11805, EY11756, EY11754, EY11755, and EY11752, Bethesda, Maryland; and by Essilor of America, Marchon Eyewear, Marco Technologies, and Welch Allyn.

 Submitted for publication November 22, 2002; revised February 20, 2003; accepted March 10, 2003.

 Disclosure: R.E. Manny, None; M. Hussein, None; J. Gwiazda, None; W. Marsh-Tootle, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Table 1.

 

Baseline Characteristics of the Sample of COMET Children in the Current Study Compared with the Entire COMET Population

	Sample of COMET Children in Current Study (n = 86)	All COMET Children (n = 469)	P *
Age (y)
Mean ± SD	9.4 ± 1.3	9.3 ± 1.3	0.51
Median (range)	10 (6–11)	9 (6–11)
Gender, % female (n)	54.7 (47)	52 (244)	0.72
Ethnicity % (n), †
White	49 (42)	46 (218)	0.73
Hispanic	22 (19)	15 (68)	0.08
Asian	6 (5)	8 (36)	0.66
Black	21 (18)	26 (123)	0.35
Other/Mixed	2 (2)	5 (24)	0.40

	OD	OS	OD	OS	OD	OS
Subjective Refraction (D), ‡
Mean ± SD	−2.35 ± 0.75	−2.35 ± 0.73	−2.33 ± 0.81	−2.32 ± 0.81	0.83	0.75
Median (range)	−2.25 (−1.00 to −4.50)	−2.25 (−1.50 to −4.50)	−2.25 (−1.00 to −4.75)	−2.25 (−0.75 to −4.625)
LogMAR baseline VA
Mean ± SD	0.003 ± 0.076	0.008 ± 0.059	0.021 ± 0.074	0.028 ± 0.069	0.04	0.02

Figure 1.

 

The signed difference in best corrected logMAR visual acuity (1 month minus baseline) versus the average of the 1-month and baseline logMAR visual acuities. Solid line: mean signed difference in logMAR visual acuity; area within the dotted lines: 95% limits of agreement between logMAR visual acuity at 1 month and baseline. (•) One child; (▪) two children; (♦) three children. Top: right eye data; bottom: left eye data.

Table 2.

 

Kappa Statistic: the Level of Agreement within 1 Line of Acuity

	n	Right Eye Kappa (95% confidence limit)	Left Eye Kappa (95% confidence limit)
Overall agreement	86	0.81 (0.69, 0.94)	0.86 (0.75, 0.97)
Female	47	0.91 (0.79, 1.00)	0.91 (0.80, 1.00)
Male	39	0.68 (0.45, 0.92)	0.80 (0.61, 0.98)
Less than 10 years	41	0.90 (0.75, 1.00)	0.80 (0.61, 0.98)
10 years or more	45	0.71 (0.50, 0.92)	0.90 (0.77, 1.00)

Table 3.

 

LogMAR VA Norms for Young Adults and Preschool Children

Reference	Subjects	Methods	Mean VA logMAR ± SD (Snellen)
Brown, Lovie-Kitchin 33	n = 10; mean age: 18 y	Monocular VA (Bailey-Lovie Chart)	−0.163 ± 0.09 6/4 (20/13)
Elliott et al. 26 (from Table 1 )	n = 37; age range: 18–24 y	Optimal refractive correction (logMAR Charts)	−0.13 ± 0.06 6/4 (20/14)
Owsley et al. 35	n = 12; age range: 19–28 y	Best optical correction (Bailey-Lovie Chart)	−0.17 6/4 (20/13)
Wood, Bullimore 36	n = 14; age range: 20–29 y	Best correction reported, right eye (Bailey-Lovie)	−0.15 ± 0.12 6/4 (20/14)
Sprague et al. 25 (from Table 6)	n = 92; age range: 3–5 y	Habitual correction not verified; right eye; full chart exposed (no isolation); HOTV and E in logMAR chart with six letters/line	0.097 6/7 (20/25)

Table 4.

 

Previous LogMAR VA Reliability Studies

Study	Subjects	Method	Significant Change in VA with 95% LOA
Blackhurst, Maguire and The Macular Photocoagulation Study Group 31	N = 82; age: ≥18 y; (from NEI website) macular disease and individuals with normal vision (?) for VA certification protocol VA: 20/20, <20/800	Best refraction Bailey-Lovie × 2; Different examiners on the same day	Recommendation in discussion, 2 lines; from Table 1 : SD = ±0.085 (converted from 0.85 MPS lines); 95% LOA = ±0.167 logMAR (±8.33 letters)
Camparini et al. 37	N = 57; age range: 21–85 y; referred for ophthalmic problems (refractive error, cataract, trauma, macular degeneration, glaucoma, diabetic retinopathy); VA: 20/100 to 20/10	Best refraction ETDRS × 4 on right eye; Standard ETDRS × 2; ETDRS Fast Procedure × 2; sessions separated by 1 to 24 hours; single examiner	From Table 2 : Standard ETDRS SD = ±0.08 logMAR; 95% LOA = ±0.157 logMAR (±7.8 letters)
Camparini et al. ARVO Abstract 2068, 2001	N = 40; age range 3–12 y	Refraction not stated; Landolt-C ETDRS-style chart; standard protocol × 2; C-Fast Procedure × 2	Standard protocol: SD = ±5.57 letters; 95% LOA = ±10.9 letters; C-fast protocol: SD = ±3.32 letters; 95% LOA = ±6.5 letters
Elliott, Sheridan 38	N = 20; mean age: 64 y; normal vision	Best refraction and unaided; Ferris logMAR × 2, 1 week apart; not clear whether data from both eyes are pooled or only dominant eye reported	Best refraction: 95% LOA = ±0.07 logMAR (±3.5 letters); unaided: 95% LOA = ±0.21 logMAR (±11 letters)
Kheterpal et al. 15	N = 18; mean age: 4.8 y (3.5–8 y); refractive and binocular disorders, three with normal vision	Unaided logMAR VA × 2, 1 week apart; matching response	OD: 95% LOA = ±0.21 logMAR; OS: 95% LOA = ±0.25 logMAR; 14 children with 1 eye consistently better; better eye 95% LOA = ±0.15 logMAR; poorer eye 95% LOA = ±0.22 logMAR
Lovie-Kitchin 39	N = 115; age range: 11–74 y; students and patients with no ocular disease	Right eye unaided; Bailey-Lovie × 2 same session; not forced to guess	Standard protocol; SD = ±0.08 logMAR; 95% LOA = ±0.16 logMAR
Lovie-Kitchin, Brown 40	N = 79; age range: 21–68 y; normal vision by self-report	Habitual correction; Bailey-Lovie × 5; 2 to 3 days apart	Recommendation in discussion, 1 line; from Table 2 : best eye Session 1 vs. 5; 95% LOA = ±0.105 logMAR (±5.25 letters)
McGraw et al. 14	N = 68; mean age: 5.32 y; subset of 116 visually normal children	Unaided; Glasgow Acuity Cards; line acuity with crowding bars; letter spacing = 0.5 letter diameter	95% LOA = ±0.10 logMAR (±4 letters)
Raasch et al. 34	N = 19; age range: 20–33 y; free of eye disease VA of 6/7 or better	Habitual VA; simulations based on probability of seeing curves generated from empirical data	From Table 1 ; British letters SD = ±0.036 logMAR; 95% LOA = ±0.07 logMAR (±3.5 letters)
Reeves et al. 32	N = 56; age range: 20–77 y; various clinical conditions with VA of at least 20/60	Best refraction; monocular Bailey-Lovie × 2; 4 to 12 weeks apart; only 1 eye reported	From Table 1 ; SD = ±0.10 logMAR; 95% LOA = ±0.19 logMAR