The measurement of visual acuity is one of the most important and frequently administered tests in professional eye care. As such, it is important to use a method that is both accurate and repeatable.
William Feinbloom was an optometric pioneer who promoted the idea that visually impaired patients required special testing methods.1 When the Feinbloom chart was introduced in the first half of the 20th century, it represented an advance in the low-vision rehabilitation field. It was the first chart to be designed specifically for use with the visually impaired and included numeral optotypes from 0 to 9 and ranging in size from 10 foot letter (3 M) to 700 foot letter (213 M). Optotypes are labeled in foot letter and (in newer versions) M notation and are not converted to Snellen (20/x) equivalent. The pages are printed on one side only and vary from one optotype per page to up to five lines per page with up to nine optotypes per line.
The high-contrast optotypes on the Feinbloom chart are calibrated by their height. For example, the height of a 40 foot letter (12 M) optotype subtends 5 min of arc at 40 feet (12.2 m) and measures 0.7 inches (18 mm). Our measurements indicated the width of the optotypes averaged about 73% of their height, except for the “1,” which was only about half of its height. We measured stroke width at 16.2% of the optotype height, or a height to stroke width ratio of 6.2.
Advantages of the Feinbloom chart include large optotype sizes, portability, and the ability to be used at any distance. Also, the use of numbers is advantageous, as many people who are not literate in English are familiar with Arabic numerals. Disadvantages include optotypes that have an unequal difficulty of identification, inconsistent width to height ratios, irregular progression of optotype sizes, and irregular crowding as one progresses through the chart.
New principles for the design of a visual acuity chart were introduced in 1976, which vastly improved the accuracy and precision of visual acuity testing.2 These ideas, with the addition of Sloan optotypes,3 evolved into what has become known as the ETDRS chart,4,5 so named for its inclusion in the Early Treatment of Diabetic Retinopathy Study. Key features of the ETDRS chart include an equal number (five) of nearly equally legible Sloan optotypes in each row, consistent spacing between optotypes and rows, and consistent 0.1 logMAR decline in optotype size as one progresses down the chart. Further guidelines for the use of this chart were clarified as recently as 1996.6 These design elements make the ETDRS chart equally valid at various testing distances, and this attribute has been independently verified.7 Since then, the ETDRS chart has been used as a gold standard against which other methods have been compared.8 It has been verified in an electronic version9 and has been shown to provide repeatable results in testing normally sighted children from 6 to 11 years old.10
Although the ETDRS chart has been readily accepted by low-vision rehabilitation practitioners, there are clinical situations when a handheld book chart, such as the Feinbloom chart, is often used. These situations may include on-site exams at nursing homes or schools for the visually impaired, where portability of the chart is important. In addition, eye care professionals who do not specialize in low-vision rehabilitation often choose a handheld book chart for its economy of price and minimal office space required. Despite the obvious advantages of the ETDRS chart, portable book charts still enjoy considerable popularity.
The LEA Numbers Low Vision Book (LNLVB) is a handheld visual acuity chart that addresses some of the disadvantages of the Feinbloom chart. The high-contrast optotypes are LEA Numbers, developed in the 1990s using a novel method of calibration to determine optotype sizes empirically by comparison with the Landolt C. To provide equal difficulty of identification, only the specially designed numerals 5, 6, 8, and 9 were used in the chart design.11
The LEA optotypes are taller than the equivalent Feinbloom optotypes (Fig. 1). By coincidence, the average of optotype height and width is nearly identical to the size predicted by 5 min of arc for the indicated distance. For example, we measured the 40 foot letter (12 M) optotypes at 21.5 mm high and 15 mm wide, for an average of 18.25 mm. A target subtending 5 min of arc at 40 feet (12.2 m) would be 17.7 mm, only 3% different from this average of height and width. Optotype sizes range from 400 foot letter (120 M) to 4 foot letter (1.2 M). The pages are printed on both sides and vary from one optotype per page to up to 12 lines per page with up to five optotypes per line. Ideally, optotype size should follow a geometric progression whereby the change in angular size is uniform from one optotype size to the next.2 The LNLVB incorporates 0.1 logMAR progression between lines or pages. Stroke width averaged 13.7% of the optotype height by our measurements, for a height to stroke width ratio of 7.3.
In addition to the advantages listed for the Feinbloom chart, the LNLVB also offers optotypes that have similar difficulty of identification, systematic 0.1 logMAR progression from one size to the next, and an included near visual acuity chart with consistent spacing and 0.1 logMAR line progression. Disadvantages of the LNLVB include a 25% chance of correctly guessing an optotype and irregular crowding, especially with the larger optotypes. However, it should be noted that, for the 50 foot letter (15 M) lines and smaller, ETDRS-spacing principles are followed. From 100 foot letter (30 M) to 63 foot letter (19 M), ETDRS-spacing principles are followed with the exception that there are only three or four optotypes per line instead of five.
Because visual acuity measurement is affected by the crowding phenomenon,12 it can be inferred that these book tests of visual acuity should produce slightly different results than the ETDRS chart based on their spacing irregularities. Both the Feinbloom and LNLVB have irregularities in character spacing, but the LNLVB is less variable. LEA optotypes 100 foot letter (30M) or smaller are in agreement with recent research that has found that spacing from 3.75 to 5 stroke widths will not cause variation in the visual acuity measurement.13
Previous studies with adult subjects found that visual acuity measurements taken with LEA Numbers correlate with those taken with Landolt C and Sloan letter charts.14,15 Although test distance will vary in our study depending on the subject’s acuity level, it is important to note that other studies on normally sighted subjects have found that LEA Numbers acuity measurements do not vary with different test distances.16 Although the Feinbloom chart has been used to measure visual acuity in a variety of studies, a search of both PubMed and Ovid identified no studies in which the accuracy or precision of the Feinbloom chart was investigated in visually impaired subjects. One study found good correlation between the Feinbloom chart, two different Snellen charts, and LEA Symbols (also calibrated against the Landolt C) in normally sighted subjects uncorrected for refractive error when visual acuity was better than 20/200.17 The results of our study should provide information on the clinical efficacy of the two charts in terms of accuracy and repeatability.
Our study was designed with two main objectives. The first objective was to determine which handheld book chart would provide visual acuity measurements more similar to the ETDRS chart. Second, we sought to determine which handheld chart provided the most repeatable measure of visual acuity. These findings can assist practitioners in determining which chart should be used in situations where the ETDRS is not available, or not feasible to use, when testing visual acuity in a visually impaired population.
This project received approval from the Northeastern State University Institutional Review Board, and subject recruitment began soon thereafter. Subject recruitment took place from two sources of visually impaired patients. The first group of subjects included visually impaired adults, 35 to 88 years old, referred to the low-vision clinic at Northeastern State University, Oklahoma College of Optometry. The second group of subjects included visually impaired students, ages 8 to 19 years, from the Oklahoma School for the Blind in Muskogee, Oklahoma. Our study included 18 participants, seven female and eleven male. The average age was 33 years (SD, 25.9 years). Each participant signed a consent form before participation in the study. If the participant was unable to read the print on the form, the form was read aloud to them before they signed. Participants <18 years of age were also required to have a signed parental assent form. To participate in the study, subjects were required to demonstrate visual acuity in the better eye between 20/80 and 20/600, according to previous measurements. All measurements were performed monocularly using only the right eye through the subject’s habitual distance-corrective lenses, unless the right eye did not meet the inclusion criteria, in which case testing was done only on the left eye.
The LNLVB and Feinbloom charts were presented initially at 10 feet (3 m). If the patient could not advance to the 200 foot letter (60 M) page, which features a minimum of three (Feinbloom) or four (LNLVB) optotypes per page, the distance was shortened to five feet (1.5 m). This protocol was used to ensure at least some crowding effect in all visual acuity measurements. When using the ETDRS chart, testing began at four meters. If the patient could not read letters on the second row, the test distance was reduced to two meters or, if required, one meter to ensure a crowding effect from the top row.
Determining the point at which to stop testing is important in measuring visual acuity. A 2001 study found that termination of testing after four of five characters were missed on a line for the ETDRS chart was sufficient.18 Because the number of optotypes per line or per page varied in the charts we tested, such a termination rule could not be used in our study. We therefore encouraged subjects to guess until they felt they could see no further. Identical instructions were used in all three charts in an attempt to standardize testing procedure. This method also gave subjects the opportunity to identify optotypes that were more easily resolved near threshold, such as the “7” on the Feinbloom chart.
Differences in scoring and testing procedure exist between the three charts used in our study, so a standard protocol was developed to facilitate statistical comparison between the three measurements. The ETDRS chart has the same number of characters (five) on each line, but the LNLVB and Feinbloom charts vary in the number of characters per line and page. A scoring procedure often used with the ETDRS chart assigns a 0.02 logMAR value for each correctly identified optotype. This allows for interpolation between rows.4 The Feinbloom and LNLVB charts are not usually used in this way, partly because the unequal number of optotypes per row or page can result in different results based on whether a large or small optotype is correctly identified. Although we recognize this as a weakness in the study, we decided it was the only way we could calculate statistical comparisons between the charts.
To reduce the error introduced by this method, we used test distances with both handheld book charts that ensured testing was terminated where there were at least three optotypes per page. We then assigned values to each character on the charts based on the represented acuity level and the number of characters on that line. For example, the 12th page of the Feinbloom chart has three characters (7, 2, and 6) and represents an acuity level of 10/100 or 1.0 logMAR when used at 10 feet (3 m). The previous page represents 10/120 or 1.08 logMAR at 10 feet (3 m). To assign a value to each of the characters on the 1.0 logMAR page, the value of each character must equal one third of the difference between 1.08 and 1.0, or 0.027 logMAR. Each correctly identified character would therefore subtract 0.027 logMAR from the previous line’s logMAR value of 1.08. So, if a subject identified all optotypes through the 1.08 logMAR page and then correctly identified two of the three 1.00 logMAR characters and none below that, the final logMAR score would be 1.03 (1.08 − 0.027 − 0.027). Using this method, we developed scoring sheets for the Feinbloom chart and LNLVB that are similar to those frequently used with the ETDRS chart.
We attempted to hold luminance values for the LNLVB and Feinbloom charts constant using identical bright room illumination for all testing. The ETDRS chart was brighter because it was backlit by LEDs in an illuminated cabinet (Good-Lite #500600). Background luminance measurements for the backlit ETDRS chart ranged from 130 to 140 cd/m2, whereas LNLVB and Feinbloom charts measured from 60 to 70 cd/m2 (Minolta Luminance Meter 1°, Minolta Camera Co. Ltd., Japan).
After the initial measurement of visual acuity, we retested the same eye under the same conditions 15 to 30 mins later. A different ETDRS chart was used for the retest. The order in which the charts were presented was randomized in both the original and second measurement of visual acuity.
We analyzed our data by regression and by Bland–Altman analysis.19 Compared with other methods, Bland–Altman is considered a very useful tool for this purpose.20 Values are presented in upper and lower Limits of Agreement, which indicate 95% confidence ranges, which were calculated as ±1.96 standard deviation of the bias. For the regression analysis, the findings are plotted, and we include the slope and intercept as well as the coefficient of determination (R2). Correlation probability (p) was calculated from the regression analysis.
The mean logMAR visual acuity for the first measurement was 1.033 for ETDRS, 1.104 for the LNLVB, and 1.110 for Feinbloom. Thus, the Feinbloom chart and the LNLVB found slightly worse visual acuity on average than the ETDRS chart. This difference was statistically significant, with paired, one-tailed t-test scores of p = 0.001 for LNLVB vs. ETDRS and p = 0.008 for Feinbloom vs. ETDRS. On each of the three charts, the mean of the second measurement showed very slightly better visual acuity than the mean of the first, but this difference was not statistically significant (paired, one-tailed t-test scores were p = 0.353 for ETDRS, p = 0.171 for LNLVB, and p = 0.350 for Feinbloom). Data gathered on each subject and the mean findings are listed in Table 1.
LogMAR acuities taken from the first round of testing for each chart were evaluated using Bland–Altman plots and regression analysis. The 95% limits of agreement for LNLVB–ETDRS were +0.099 and −0.240 (Fig. 2). The scatter plot slope was 0.880, the intercept was 0.062, the coefficient of determination (R2) was 0.908, and the correlation coefficient (R) was 0.953 (p < 0.0001) (Fig. 3). The Bland–Altman analysis for Feinbloom–ETDRS showed the 95% limits of agreement of +0.169 and −0.322 (Fig. 4). Regression analysis indicated a slope of 0.809 and intercept of 0.135 with R2 = 0.818 and r = 0.905 (p < 0.0001) (Fig. 5). Thus, the LNLVB showed slightly better agreement with ETDRS than Feinbloom.
On test–retest comparisons, the ETDRS showed the highest repeatability, with the 95% limits of agreement of +0.117 and −0.128 (Fig. 6A). A scatter plot of the two sets of data showed a slope of 0.911 and intercept of 0.097 with R2 = 0.952 and r = 0.976 (p < 0.0001). The test–retest 95% limits of agreement for the LNLVB and Feinbloom were very similar at +0.159/−0.200 (Fig. 6B) and +0.184/−0.202 (Fig. 6C), respectively. For the LNLVB, the scatter plot slope was 0.892, the intercept was 0.138, R2 = 0.910, and the r = 0.954 (p < 0.0001). The scatter plot slope for the Feinbloom chart was 1.026, the intercept = −0.020, R2 = 0.888, and r = 0.942 (p < 0.0001).
The LNLVB chart showed closer correlation with ETDRS than Feinbloom, which is not surprising considering the design advantages the LNLVB has over the Feinbloom chart. To interpret the data, it is useful to consider the clinical significance of the Bland–Altman analysis results. The limits of agreement for the Feinbloom when compared with the ETDRS signified that 95% of the time the Feinbloom logMAR score would fall within a range of +0.169 and −0.322 logMAR of the ETDRS measurement. In Snellen format, this would indicate that if a patient had an ETDRS visual acuity of 20/216 (the average of our visually impaired subjects), you could be 95% confident that the Feinbloom acuity of that subject would fall between 20/103 and 20/318. For the LNLVB, the 95% confidence range would be between 20/124 and 20/271.
Our second objective was to determine which visual acuity chart would produce the most repeatable results. The ETDRS chart proved to be the most repeatable test in terms of agreement and correlation between the first and second measurements, which was expected, considering it is the gold standard. The Feinbloom and LNLVB were slightly less repeatable but extremely comparable with each other in this regard. In Snellen terms, given an initial ETDRS visual acuity of 20/216, the ETDRS would have a 95% confidence interval from 20/161 to 20/283. Feinbloom would have a 95% confidence interval from 20/136 to 20/330 and LNLVB would be 20/136 to 20/311, given the same initial visual acuity of 20/216.
Visual acuity measurement is more variable as visual acuity becomes worse, as shown in a 2005 study where the standard deviation of measurement error was 0.04 logMAR in subjects with decimal visual acuity of 0.7 (20/29) or better and 0.09 logMAR in subjects with decimal visual acuity of 0.3 (20/67) to 0.45 (20/44).21 One might expect even more variability in our more visually impaired population. Other studies involving normally sighted children10 and adults tested under ideal conditions22 showed visual acuities would have to vary by ±0.14 or ±0.15 logMAR, respectively, to be clinically significant at 95% limits of agreement. Our ETDRS findings were within this range (+0.117/−0.128). LNLVB (+0.159/−0.200) and Feinbloom (+0.184/−0.202) results were only slightly outside this range, even though we were testing subjects with significant visual impairment.
It has been known for nearly a century that visual acuity measurements are affected by luminance.23 Our luminance values varied between the handheld charts and the backlit ETDRS chart. Although this could be considered a weakness of our study design, the charts were each used as they are generally used in a clinical setting. Luminance values were identical for LNLVB and the Feinbloom chart, which were the tests of primary interest. The greater luminance of the ETDRS chart may explain, at least in part, why this chart showed better visual acuity than that found with the Feinbloom chart or LNLVB. The relationship between the log of luminance and logMAR visual acuity is considered a linear function and, over a “normal” photopic range of 40 and 600 cd/m2, a doubling of luminance has been shown to result in one additional correctly identified optotype on a chart with five optotypes per row.24 In our case, the ETDRS chart was about double the luminance of the other two charts. Using this formula, the ETDRS should have measured about 0.02 logMAR units better than the other two charts. The actual difference was closer to 0.08 logMAR units, so perhaps luminance was not the only factor. Other studies7,25 have shown ETDRS to result in significantly better visual acuity than Snellen charts, which suffer from some of the same irregularities in spacing and crowding as the Feinbloom chart and LNLVB.
Of note anecdotally was the ease of identification of the “7” optotype on the Feinbloom chart. After both rounds were complete, subjects were asked to select the optotype that was easiest to identify on the LNLVB and Feinbloom chart. For the Feinbloom chart, nearly all subjects chose the “7” optotype. Of the 10 Arabic numerals, only the “1” has been shown to be more identifiable than the “7.”26 In the Feinbloom chart, the “1” does not appear until the 40 foot letter (12M) row, so many of our subjects did not see it. The “7” appears five times in sizes larger than 40 foot letter (12 M). No preference was shown to any LEA Number optotype as being more easily identified. The numerals chosen as LEA Number optotypes (5, 6, 8, and 9) are the four most difficult numerals to identify.26
The LNLVB is available with binding on the top (Good-Lite #527300), like the Feinbloom chart, or on the left side (Good-Lite #513100). Because the LNLVB is printed on both sides of each page, we noticed it was easier to progress in order when using the chart with the binding on the left side.
The sizes of optotypes in the LNLVB were designed empirically rather than being based on a 5 min of arc height corresponding to the labeled distance. For this reason, some people do not accept this test. The height of the LEA Numbers is slightly taller than an equivalently labeled Feinbloom optotype (Fig. 1). Two factors might explain why the two charts provide similar results despite the size difference in the optotypes. For maximum legibility, the height of an optotype should be from four to six times the stroke width.26 The Feinbloom optotype heights are 6.2 times the stroke width, whereas the same measurement on the LEA Numbers is 7.3. This would decrease the ability to successfully identify optotypes when using LEA Numbers compared with Feinbloom optotypes. Second, as previously stated, the LEA Numbers use only the four numerals that are most difficult to identify.26
Overall, the results of our study indicate that there is a slight advantage, in terms of agreement with ETDRS, of LNLVB over the Feinbloom chart when testing visually impaired patients. The two tests proved nearly identical in terms of repeatability in visually impaired subjects. Both handheld charts measured slightly worse visual acuity than the ETDRS chart, but overall proved to be suitable alternatives when the ETDRS chart is not available or practical in certain clinical situations.
Our project included a small sample, typical of a pilot study. Further research using a larger sample size is recommended.
Northeastern State University
Oklahoma College of Optometry
1001 N Grand Avenue
Tahlequah, Oklahoma 74464
None of the authors have any financial or other conflict of interest in the products we evaluated in this study. This study was made possible by funds obtained from Northeastern State University Oklahoma College of Optometry. We thank Richard Hoenes and Thomas Salmon, OD, PhD, for their help in analyzing the data for our study. This study was presented as a poster at the meeting of the American Academy of Optometry, Boston, MA, October 12, 2011.
Received December 12, 2011; accepted June 8, 2012.
1. Feinbloom W. Introduction to the principles and practice of sub-normal vision correction. J Am Optom Assoc 1935; 6: 3–18.
2. Bailey IL, Lovie JE. New design principles for visual acuity letter charts. Am J Optom Physiol Opt 1976; 53: 740–5.
3. Sloan LL, Rowland WM, Altman A. Comparison of three types of test target for the measurement of visual acuity. Q Rev Ophthalmol 1952; 8: 4–16.
4. Ferris FL III, Kassoff A, Bresnick GH, Bailey I. New visual acuity charts for clinical research. Am J Ophthalmol 1982; 94: 91–6.
5. Ferris FL III, Sperduto RD. Standardized illumination for visual acuity testing in clinical research. Am J Ophthalmol 1982; 94: 97–8.
6. Ferris FL III, Bailey I. Standardizing the measurement of visual acuity for clinical research studies: guidelines from the Eye Care Technology Forum. Ophthalmology 1996; 103: 181–2.
7. Kaiser PK. Prospective evaluation of visual acuity assessment: a comparison of snellen versus ETDRS charts in clinical practice (An AOS Thesis). Trans Am Ophthalmol Soc 2009; 107: 311–24.
8. Ruamviboonsuk P, Tiensuwan M, Kunawut C, Masayaanon P. Repeatability of an automated Landolt C test, compared with the early treatment of diabetic retinopathy study (ETDRS) chart testing. Am J Ophthalmol 2003; 136: 662–9.
9. Cotter SA, Chu RH, Chandler DL, Beck RW, Holmes JM, Rice ML, Hertle RW, Birch EE, Moke PS. Reliability of the electronic early treatment diabetic retinopathy study testing protocol in children 7 to <13 years old. Am J Ophthalmol 2003; 136: 655–61.
10. Manny RE, Hussein M, Gwiazda J, Marsh-Tootle W. Repeatability of ETDRS visual acuity in children. Invest Ophthalmol Vis Sci 2003; 44: 3294–300.
12. Flom MC, Weymouth FW, Kahneman D. Visual resolution and contour interaction. J Opt Soc Am 1963; 53: 1026–32.
13. Shah N, Laidlaw DA, Brown G, Robson C. Effect of letter separation on computerised visual acuity measurements: comparison with the gold standard Early Treatment Diabetic Retinopathy Study (ETDRS) chart. Ophthalmic Physiol Opt 2010; 30: 200–3.
14. Johnson CA, Lyons S, Quinn N, Hartzell J, Travison T, Suckow M, Huffer S, Loussifova M, Moore B. Measurement of visual acuity with LEA Symbols and LEA Numbers: comparison to adult standards. Invest Ophthalmol Vis Sci 2005; 46:E-abstract 1934.
15. Vaidhyan JJ, Lietzen P, Ah-Kine Ng Poon Hing D, Pathak A, Ahonen E, Quinn N, Lyons S, Leinonen M, Hyvarinen L, Moore B. Comparison of visual acuity measured with LEA Symbols and LEA Numbers to adult standards. Invest Ophthalmol Vis Sci 2007; 48:E-abstract 4853.
16. Ah-Kine Ng Poon Hing D, Vaidhyan JJ, Pathak A, Quinn N, Deng L, Lyons S, Moore B. Comparison of visual acuity measured with LEA Symbols and LEA Numbers at different test distances. Invest Ophthalmol Vis Sci 2007; 48:E-abstract 4852.
17. Snyder E, Pham C, Steele G, Maino D, Caden B. A comparison of four visual acuity tests. Optom Vis Sci 1999; 76 (suppl.): 155.
18. Carkeet A. Modeling logMAR visual acuity scores: effects of termination rules and alternative forced-choice options. Optom Vis Sci 2001; 78: 529–38.
19. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986; 1: 307–10.
20. Raasch TW, Bailey IL, Bullimore MA. Repeatability of visual acuity measurement. Optom Vis Sci 1998; 75: 342–8.
21. Leinonen J, Laakkonen E, Laatikainen L. Random measurement error in visual acuity measurement in clinical settings. Acta Ophthalmol Scand 2005; 83: 328–32.
22. Arditi A, Cagenello R. On the statistical reliability of letter-chart visual acuity measurements. Invest Ophthalmol Vis Sci 1993; 34: 120–9.
23. Ferree CE, Rand G. The effect of intensity of illumination on the acuity of the normal eye and eyes slightly defective as to refraction. Am J Psychol 1923; 34: 244–9.
24. Sheedy JE, Bailey IL, Raasch TW. Visual acuity and chart luminance. Am J Optom Physiol Opt 1984; 61: 595–600.
25. Rosser DA, Cousens SN, Murdoch IE, Fitzke FW, Laidlaw DA. How sensitive to clinical change are ETDRS logMAR visual acuity measurements? Invest Ophthalmol Vis Sci 2003; 44: 3278–81.
26. Kuntz JE, Sleight RB. Legibility of numerals: the optimal ratio of height to width of stroke. Am J Psychol 1950; 63: 567–75.