Optotypes

Most familiar acuity tests require the subject to identify letters arrayed in rows of decreasing size. The so-called Snellen acuity test is the prime example, although Snellen acuity now usually refers to a way of reporting test results rather than to any particular type of chart. To facilitate testing of young children and people unfamiliar with the Latin alphabet, other optotype tests based on the tumbling E, Landolt C, numerals, or simple pictures of familiar objects are also used. Visual acuity can also be assessed with grating patterns, but grating acuity often overestimates Snellen acuity in patients with age-related maculopathy,³ typically by a factor of 2 or more.

The NAS-NRC recommends that the Landolt C test be used as the standard by which other acuity tests are compared. The Landolt C is a broken circle in which the width of the break and the stroke width are both equal to one-fifth the height of the C. The ring is shown with the break at one of four locations, and the subject responds by saying “right,” “down,” “up,” or “left,” or by pointing in the appropriate direction.

There are several advantages to the Landolt C test, including equal difficulty of all targets (unlike letters that vary in degree of difficulty), sensitivity to astigmatic refractive error, and suitability for use with illiterate subjects. However, the Landolt C test is not widely used because it has a guessing rate of 25%, so an alternative is to standardize another optotype set by comparing acuities obtained with it to Landolt C test acuities. The Sloan letters,⁴ a set of 10 upper-case sans serif letters, are the most popular substitute. An acuity chart based on Sloan letters was developed for the Early Treatment Diabetic Retinopathy Study (ETDRS)⁵ and is illustrated in Fig. 11.1. The original ETDRS chart has been replaced by the 2000 series revised ETDRS chart that more accurately equates the difficulty of letters on all lines. The ETDRS charts are the most widely used acuity charts for clinical research.

Fig. 11.1 The Early Treatment Diabetic Retinopathy Study (ETDRS) acuity chart.

(Reproduced with permission from Ferris FL III, Kassof PA, Bresnick GH, et al. New visual acuity charts for clinical research. Am J Ophthalmol 1982;94:91–96.)

Chart layout

Careful attention must be paid to the layout of the acuity chart. The chart should follow a uniform progression of letter sizes, typically a 0.1 log unit (or 26%) reduction in size from line to line. The uniform progression ensures that a one-line loss will have the same meaning at any point on the chart and at any viewing distance. The same number of letters should appear on each line, and the spacing should be uniform, both within and between lines. The spacing requirement results in a large chart, with the letters forming an inverted triangle. The NAS-NRC recommends 8–10 letters per line, but studies⁶ suggest that as few as three letters are required for an accurate estimate of visual function. The ETDRS chart uses five letters per line.

Concern about uncontrolled “crowding” effects has led to further modifications of the ETDRS chart. Crowding refers to the reduced visibility of letters when they are surrounded by other letters. Crowding has a larger effect in some types of visual dysfunction, notably amblyopia⁷ and macular degeneration.⁸ For most acuity charts, letters at either the end of a line or at the top or bottom of the chart are less subject to crowding than are internal letters. To equalize crowding effects, contour interaction bars may be added around the perimeter of the chart.⁹

Acuity test distance

ETDRS charts are available for a range of test distances from 4 meters (13 feet) to 2 meters (6.5 feet) and when used at the designated distance can measure acuities from 20/10 to 20/200. However, given the logarithmic progression of letter sizes, they can be used at any distance with an appropriate correction of the reported results. For patients with very poor acuity, the clinician may resort to finger counting or hand motion. This strategy is strongly discouraged by low-vision practitioners because it can be demeaning and depressing for the patient to be left with the impression that their vision is so poor it cannot even be measured with an eye chart. Instead, it is recommended that the patient be moved closer to the chart. Using a test distance of 50 cm and appropriate refractive correction it is possible to reliably measure “counting finger” acuity with an ETDRS chart (approximately 20/1460 ± 10%).¹⁰ “Hand motions” can be measured with some electronic acuity tests (see below). Distance itself should have little effect on visual acuity, provided that the subject’s accommodative state and pupil size are controlled. However, one study¹¹ found that acuity changed by as much as seven letters (more than one ETDRS line) when the test distance was reduced to less than 2 meters. The reason for this discrepancy remains unexplained.

Luminance and contrast

Whatever the test distance, the chart must be adequately illuminated and of high contrast. Illumination standards vary from 100 cd/m² in the USA to 300 cd/m² in Germany. Increasing chart luminance improves visual acuity in normal subjects, but reaches a plateau at about 200 cd/m².¹² Various types of visual dysfunction can change the effects of luminance on acuity. For example, patients with retinitis pigmentosa may show a decrease in acuity at higher luminance, whereas patients with age-related macular degeneration often continue to improve at luminances well above the normal plateau.¹³ A luminance standard of 100 cd/m² can be justified because it represents good room illumination for ordinary reading material. Furthermore, most of the currently proposed standards would yield the same acuity scores, plus or minus one letter (assuming five letters per line and a 0.1 log unit size progression) in normal subjects.

The relationship of visual acuity to letter contrast follows a square-root law.¹⁴ For example, decreasing contrast by a factor of 2 would decrease acuity by roughly a factor of 1.4. The NAS-NRC recommends that letter contrast be at least 0.85. Transilluminated, projection, and reflective charts (wall charts) can all meet these standards, but some transilluminated charts are deficient in luminance, and some projection systems lack sufficient luminance and contrast. Accurate calibration requires a spot photometer, for which procedures are described in the NAS-NRC document.¹

Test administration

Administration of visual acuity tests is simple and straightforward. However, one detail often overlooked in clinical testing is that the test must be administered in a “forced-choice” manner. Rather than allowing the patient to decide when the letters become indistinguishable, the patient should be required to guess the identity of each letter until a sufficient number of errors are made to justify terminating the test. People differ in their willingness to respond to questions when they are not confident about the answers. A person with a conservative criterion answers only when absolutely certain about the identity of the letter, whereas a person with a liberal criterion ventures a guess for any letter that is even barely discernible. These two people may receive different acuity scores because of differences in their criteria rather than because of variations in visual function. This is not merely a theoretical concern. Several studies^15,16 have shown that criterion-dependent test procedures lead to inaccurate and unreliable test results. Forced-choice procedures are criterion-free because the examiner, rather than the observer, determines whether the letter is correctly identified.

Scoring

Until recently, visual acuity tests were usually scored line by line with the patient being given credit for a line when a criterion number of letters were identified correctly. The NAS-NRC recommends that at least two-thirds of the letters on a line be correctly identified to qualify for passing. Allowing a small proportion of errors improves test reliability.¹⁷ For tests that follow the recommended format of an equal number of letters on each line and a constant progression of letter sizes, it is preferable to give partial credit for each letter correctly identified. This is commonly done by counting the number of letters read correctly on the entire chart and converting this to an acuity score by means of a simple formula that values each letter as L/N, where L = difference in acuity between adjacent lines and N = number of letters per line. So for a chart with five letters per line and a 0.1 logMAR (see below) progression from line to line (such as the standard ETDRS chart) each correct letter is worth 0.1/5 = 0.02 logMAR. Although differences between scoring methods are usually small, it has been shown^2,6,18 that letter-by-letter scoring is more reproducible than line-by-line scoring.

The most familiar method for reporting visual acuities is the Snellen fraction. The numerator of the Snellen fraction indicates the test distance, and the denominator indicates the relative size of the letter, usually in terms of the distance at which the stroke width would subtend a visual angle of 1 minute. Thus “20/40” indicates that the actual test distance was 20 feet and that the strokes of the letters would subtend 1 minute of arc at 40 feet.

To simplify comparison of acuities measured at different distances, the minimum angle of resolution (MAR) should be used. The MAR is the visual angle corresponding to stroke width in minutes of arc and is equal to the reciprocal of the Snellen fraction. Visual acuities are frequently converted to log10 and reported as “logMAR.” For example, 20/20 acuity corresponds to a MAR of 1 minute of arc, or a logMAR of 0, and 20/100 acuity corresponds to a MAR of 5 minutes of arc, or a logMAR of 0.7 (as does an acuity of 2/10 or 6/30). The MAR and logMAR scales increase with worsening acuity. One often sees visual acuities reported as the decimal equivalent of the Snellen fraction. However acuities are more normally distributed when converted to logMAR rather than decimal values. Moreover, decimals give the false impression that acuity scores can be equated with overall loss of visual function. For example, an acuity of 0.1 (the decimal equivalent of 20/200) may misleadingly suggest that the patient has retained 10% of residual visual function.

Specifying letter size

As with all acuity tests, the most critical parameter is the visual angle subtended by the optotype. Many systems have been devised for specification of print size. One of the most common is the Jaeger J notation. Jaeger notation is based on a numeric scale (J1, J2, and so on) that follows no logical progression except that larger numbers correspond to larger print sizes. Furthermore, print with the same J specification can vary by as much as 90% from one test manufacturer to another.²⁰

Alternatives to the Jaeger notation are the typesetter’s point system, the British N system, and the M notation introduced by Sloan. The typesetter’s point is commonly used to specify letter size for printed text and is equal to 0.32 mm (1/72 inch). However, the measurement refers to the size of the metal slug that contains the letter and varies from one font to another. A study²¹ of the effect of font on reading speed showed that the nominal sizes of printed text can be very misleading. Of four fonts, all labeled as 12 point, one was much more legible than the other three. But it turned out that the more legible font was actually larger than the others and when equated for real size there was no difference in legibility.

The British N system standardized the point size specification by adopting the Times Roman font. Sloan’s M notation,⁴ widely used in the USA, is standardized according to the height of a lower-case “x.” A lower-case 1M letter subtends 5 minarc at a 1-meter viewing distance and corresponds roughly to the size of ordinary newsprint. None of the print size specifications can be used for quantitative comparisons unless viewing distance is also specified. For example, 1M print read at a distance of 40 cm would be recorded as 0.40/1.00M.

Words versus continuous text

One issue that remains unresolved is whether near acuity tests should be based on unrelated words or meaningful text. An argument in favor of unrelated words is that contextual information promotes guessing and may lead to an overestimate of near acuity.²² In addition, the presence of semantic context introduces variability because of cognitive intellectual factors that may mask the visual factors of primary concern.²³ On the other side it is argued that the main reason for measuring near acuity is to gain information about reading performance. Since context is normally available to the reader, a reading test that includes meaningful text will be a better indicator of everyday performance. Fortunately, reading speed for meaningful text is highly correlated with reading speed for unrelated words.²⁴

The MNREAD test²⁵ uses meaningful text to evaluate near acuity. The test is composed of 19 standardized sentences in a logarithmic progression of sizes. Each sentence is 55 characters and the words are drawn from a controlled vocabulary. The test can be used to measure reading acuity (the smallest print size that can be read), maximum reading speed, and critical print size (the smallest print size for maximum reading speed). The test–retest variability for the MNREAD test has been assessed in patients with macular disease²⁶ and the coefficient of repeatability was found to be greater than 65 words/minute. The high level of variability may be due, in part, to the short sentences. The International Reading Speed Test (IReST)²⁷ uses 150-word paragraphs of continuous text to measure reading speed, which should yield more reproducible measures. The IReST is available in 17 languages.

Common contrast sensitivity tests

Traditional methods for measuring contrast sensitivity require relatively expensive and sophisticated equipment – typically a computer-controlled video monitor – and employ time-consuming psychophysical procedures. However, several simpler contrast sensitivity tests have been developed primarily for clinical use. These include the Functional Acuity Contrast Test⁴⁴ (FACT, replacement for the popular Vistech VCTS chart) and the CSV-1000,⁴⁵ sine-wave grating tests in chart form, and various low-contrast optotype tests such as the Lea test,⁴⁶ the Pelli–Robson letter chart,¹⁷ the Melbourne Edge Test⁴⁷ and the Mars Letter Contrast Sensitivity Test.⁴⁸ Examples of three of the most commonly used clinical contrast sensitivity tests are illustrated in Fig. 11.2.

Fig. 11.2 Commonly used clinical contrast sensitivity tests. (A) Vistech VCTS 6500.

(Courtesy of Vistech Consultants, Dayton, OH.) (B) Lea numbers low-contrast acuity test. (Courtesy of Good-Lite, Streamwood, IL.) (C) Pelli–Robson letter sensitivity chart. (Reproduced with permission from Pelli DG, Robson JG, Wilkins AJ. The design of a new letter chart for measuring contrast sensitivity. Clin Vis Sci 1988;2:187–200.)

Gratings versus optotypes

A thorough discussion of the relative merits of the various tests is beyond the scope of this chapter; however, a few salient points are worth noting. Various investigators disagree about whether measurement of an entire CSF is necessary or whether a single measure of contrast sensitivity is adequate for clinical purposes. Proponents of the sine-wave grating tests argue that visual dysfunction can cause reductions in contrast sensitivity over a limited range of spatial frequencies, which would be missed by more global measures of contrast sensitivity. On the other hand, advocates of global measures note that contrast sensitivities at specific spatial frequencies tend to be highly correlated with one another, and they maintain that overall changes in contrast sensitivity are clinically more important than subtle bumps and wiggles in the CSF. Data from large-scale studies of vision impairment indicate that global measures of contrast sensitivity are valuable predictors of difficulty in everyday life.³⁹

Test design and procedure

In order to be useful, contrast sensitivity must be measured accurately and reliably. Many of the same principles of test design previously discussed for visual acuity tests apply to contrast sensitivity tests. Most important, the test should employ a criterion-free procedure, a uniform progression of contrasts, and an adequate number of trials at each contrast to make a reliable estimate of sensitivity. Most of the optotype tests conform to good design principles and produce reliable results, with the Mars test outperforming the very popular Pelli–Robson test.⁴⁹ Sine-wave grating charts tend to be less reliable because they have a limited number of trials to make measurements at several spatial frequencies.⁵⁰

Some of the electronic acuity testing systems can also measure contrast sensitivity, including the Vision Test 2000 and FrACT. However, it is difficult to display a wide-enough range of contrasts to measure normal thresholds, and accurate calibration of the display monitor is critical. One study found that the computer-based test was less reliable than paper charts,⁴⁹ presumably due to problems generating low-contrast patterns with LCD displays.

Interpretation of clinical versus statistical significance: an example from the literature

One of the vexing problems with contrast sensitivity testing is how to interpret the clinical significance of test scores. After many decades of acuity testing, we have arrived at a consensus that a doubling of the MAR (increase of 0.3 logMAR or 15 ETDRS letters) represents a meaningful change in acuity. Recent data from large population-based studies suggest that a doubling of contrast threshold (reducing sensitivity by 0.3 logCS units or six letters on the Pelli–Robson chart) has a comparable impact on task performance and quality of life.^36,39

After several decades of laboratory investigation, contrast sensitivity testing is taking its place alongside visual acuity in clinical vision research. While early claims that contrast sensitivity would replace visual acuity proved to be exaggerated, this additional test does have an important role to play. Contrast sensitivity may not be particularly useful for differential diagnosis and screening, but its close association with everyday task difficulty has made it an important outcome measure for assessing treatment safety and efficacy.