3
VISUAL TASK PERFORMANCE
This chapter is concerned with the relationship of vision to the performance of everyday life and work tasks. In approaching the question of how to determine if a worker has a disability for visual reasons the Social Security Administration (SSA) currently relies primarily on tests of basic, objectively testable visual functions— namely, acuity and visual fields—with the implicit assumption that these measures can predict ability to perform visually intensive work and daily life tasks. As understanding of the complexity of the relationship between the traditional measures of visual function and an individual’s actual abilities to perform important tasks has grown, SSA has become concerned about the predictive validity of such tests and specifically requested the committee to explore the possibility of using measures that more directly test the ability to perform daily life and work tasks.
The use of such measures would be contingent on meeting several challenges. First is the need to select a manageable set of surrogate tasks that adequately represent important vision-related tasks of everyday life and work. Second, SSA would have to ensure that tests of such tasks proposed for use in disability determination demonstrate construct validity and are reliable and well normed. The tests also
must show a robust relationship to the requirements of jobs in the U.S. economy (i.e., must demonstrate strong predictive validity) as demonstrated in rigorous peer-reviewed research. The development of such tests is thus dependent on knowledge of the vision-related task requirements of the immense variety of jobs in the economy—jobs that are not stable but change over time.
The committee conducted several interrelated inquiries in its effort to assess (1) the relationships between visual function measures and performance of everyday life and job tasks and (2) the possibility of using tests of vision-related tasks in disability determination. First, an iterative process was used to select a limited number of task domains to represent the broad range of vision-related everyday life and work tasks. Many candidate task domains were discussed, and a short list was developed, based on the collective expertise of committee members. In a concurrent effort described below, the relationships of the most common standard vision measures to actual performance on various jobs and tasks were analyzed. The information gathered in this review, when fed back to the selection of task domains, supported the committee’s selection of four task domains as representative and inclusive of the important tasks of everyday life and work:
-
reading and related close work, such as use of a computer display,
-
mobility, both ambulatory and driving,
-
social participation, and
-
tool use and manipulative tasks.
The committee sought evidence to demonstrate (through both direct observation and self-report) the nature of the relationships between visual function measures such as acuity, visual fields, contrast sensitivity, and others and the actual ability of individuals in a community setting to perform important daily life tasks and other vision-related tasks that would be of importance in a job setting. Such evidence would scientifically support the claim that the testing of visual functions can predict job performance abilities—a predictive relationship that has not to our knowledge been rigorously demonstrated for the tests now used by SSA. We surveyed and
evaluated the experimental literature on these relationships. In addition, we conducted a concurrent literature review and analysis of the relationship of the vision measures now used by SSA—visual acuity and visual field loss—to individual self-reported functioning and health-related quality of life (HRQOL), in part because SSA requested that we evaluate the usefulness of HRQOL measures in disability determination methodology. This work provided valuable information for the committee’s task domain selection.
An important goal of this work was to characterize the form of any relationships that were found between specific vision test results and task performance. For example, is there a step function or threshold in acuity or visual field scores at which performance on particular tasks deteriorates significantly, or is the relationship relatively continuous and uninflected, meaning that a “natural” cutoff point for disability could not be derived from the relationship?
Finally, we assessed job analysis databases, two of them in depth, to determine the availability and quality of information about how important vision functions are related to the performance of job tasks, using current Department of Labor job taxonomies. In addition, we analyzed the importance of vision to specific job tasks or skills independent of the job categories. This effort was designed to evaluate the evidence available to support more specific and tailored job disability assessments that would require determination of which visual measures should be weighed in determining vision-related disability for specific job categories and the underlying job tasks in each category.
This chapter discusses the committee’s findings from this set of investigations and analyses. First we present the findings on the importance and relationships to visual functions of each of the four task domains: reading, mobility, social participation, and tool use and manipulative tasks. These sections include reviews of available tests for these task domains and recommendations on the use of such tests for disability determination. Next we present findings on health-related quality of life measures, followed by a review of the evidence from occupational analysis databases. Finally, we summarize our recommendations regarding use of tests of visual task performance in disability determinations.
READING
Description
The Case for Including Reading Performance in Disability Determination
Disability is one of the four levels of evaluation in the widely used classification of vision outlined by Colenbrander (International Society for Low Vision Research and Rehabilitation, 1999): disorder, impairment, disability, and handicap. Reading difficulty is the primary exemplar of a disability. Reading is a skill or ability possessed by a person rather than a characteristic of an eye or visual system per se.
Like other important abilities of daily life, reading performance depends in complex ways on the interaction of visual input, cognitive proficiency, oculomotor control, and probably other factors. Two people with identical eye disorders and levels of impairment (as measured by such clinical tests as acuity) might easily perform differently in reading. The same person with no change in eye disorder or measured impairment might improve his or her reading performance as a result of rehabilitation. Reading also differs from clinical tests of visual impairment in being dynamic in character. Reading, like driving and hand-eye coordination, involves rapid visual information processing and online integration of vision with other processes. As described below, impairments in acuity, contrast sensitivity, and field status (especially central field loss) affect reading performance in characteristic ways, but they do not provide accurate predictions of reading ability.
Agencies concerned with assessing disability in relation to work seek to determine whether individuals possess the capacities or skills to perform jobs. If reading is essential for a job, then a necessary condition for employment is the ability to read, with measured values of acuity or field being of only indirect relevance. Since reading is an essential component of many jobs in the modern American economy, a person with reduced reading ability is certainly disadvantaged. In this case, direct measurement of reading performance would be valuable in disability determination.
Reading is only one of many possible skills that might figure in disability assessment. How important is reading? Throughout the world, reading is one of the most highly valued activities in human culture. The United Nations and other international bodies use literacy rates as one of the primary indicators of social and economic development. To quote from Hamadache (1990): “The struggle for literacy is also a struggle for justice, for access to knowledge, and for equality. Literacy is an essential precondition for the effective exercise of human rights.” When eye disorders deprive or limit people’s access to the printed word, the issue is vision disability, not literacy, but the individual consequences may be just as severe. Because of the fundamental importance of reading, low vision is sometimes defined as the inability to read a newspaper at a normal distance with best correction (glasses or contact lenses).
Reading difficulty is often cited as the most common presenting symptom in low vision clinics. For instance, Elliott et al. (1997) reported that reading was the primary objective for 75 percent of elderly patients seeking low vision rehabilitation and the secondary objective for 21 percent. Leat et al. (1999) reported that surveyed patients had a priority for reading medicine bottles and bank statements over reading the newspaper.
During the public forum held by the committee as part of its information-gathering activities, most of the presenters noted the limitations of impairment measures for assessing disability in skills of daily life, including reading. This concern with reading is consistent with our analysis, based on the Position Analysis Questionnaire, a proprietary job analysis system, that indicated that written communication is important in 47 percent of jobs.
Given the importance of reading to employment and other activities of daily life and the imprecision of estimating reading disability from traditional measures of visual impairment, a strong case can be made for including evaluation of reading performance in disability assessment. As described below, although reading performance is related to measures of visual impairment (acuity, contrast sensitivity, and field), a direct measurement of reading performance on a well-
standardized test would provide additional information that may be relevant to disability determination.
Before presenting a recommendation about the inclusion of a test of reading in disability assessment, we address several questions: What is the range of reading tasks under consideration? What aspects of reading have been measured? What clinical tests of reading vision already exist? What are the desirable characteristics of a suitable reading test for disability assessment? After addressing these questions, we turn to an evaluation of reading tests for assessing disability.
The Range of Reading Tasks
Reading is a cluster of different tasks that impose different demands on vision, motor control and language skills. In conventional reading, people navigate through the text with a series of brief eye movements called saccades, separated by pauses called fixations. People with mild to moderate visual impairment may use optical magnifiers, requiring coordination of hand movements, head movements, and eye movements. People with more severe visual impairment may use electronic magnifiers such as closed circuit TV.
There is a trade-off between the amount of magnification and the proportion of a printed page that is visible in the magnifier’s field of view. Two problems result from the restricted field. First, there is the question of the number of letters in the field (window size) necessary to support the fastest reading. The critical number is at least four and can be greater, depending on the motor demands for magnifier scanning (cf. Beckmann & Legge, 1996; Legge, Pelli, et al., 1985). Second, the diminished field seen through the magnifier hides the global layout information on the page (Den Brinker & Beek, 1996).
Knowledge of the global layout of text is not very important for the sequential line-by-line reading of continuous prose. But nonsequential reading occurs in skimming text for gist or searching text for critical terms or hyperlinks. While the impacts of magnification and field size are fairly well understood for sequential text reading, relatively little is known about their impact on nonsequential reading.
Nonsequential reading is taking on greater importance with the growing use of computers at work and in the home. Screen magnification on computers has been implemented in commercially available software applications. Although these forms of adaptive technology have proven remarkably useful, they have generated new problems in dealing with the nonsequential display of information on computers.1 It is expected that the growing use of hypertext in computer reading and increased reliance on graphical displays (typically nonsequential in their use) will impose extra burdens on reading by people with disabilities. This is an area in need of research.
It should also be noted that researchers have experimented with nontraditional methods for displaying text in hopes of finding a method particularly advantageous for people with low vision. For instance, the RSVP method (rapid serial visual presentation), involves displaying words of a text sequentially at the same place on a display screen, minimizing the need for eye movements. People with normal vision can read RSVP text much faster than conventional text (Rubin & Turano, 1992). Despite high hopes for a similar improvement in speed for people with low vision, RSVP provides only a modest benefit for readers with low vision (Fine & Peli, 1995; Harland et al., 1998; Rubin & Turano, 1994). A variant of RSVP, called ESP (elicited sequential presentation) provides a modest benefit in reading speed (Arditi, 1999). One disadvantage of RSVP and ESP is that they do away with global layout information altogether. Reading in the real world is not restricted to books, sheets of paper, and computer screens. Signage is important for mobility, both walking and driving. Spectacle-mounted or hand-held telescopes can be useful for finding signs, but the targeting process is time-consuming and is difficult to accomplish while the viewer is in motion. Stabilization of features in the magnified retinal image requires recalibration of the relationship
between head movements and compensatory eye movements, known as the visual vestibulo-ocular reflex (Demer et al., 1988). In short, while telescopes can be used for reading and other high-acuity distance tasks, there are serious practical limitations.
Other common reading tasks in the workplace include monitoring of dials and instrument panels, retail labels, and financial documents, including currency notes (National Materials Advisory Board & National Research Council, 1995).
It is important to keep in mind that many visually disabled people use nonvisual methods for reading in addition to or instead of print. Text can be read aloud by live assistants or recorded on audio tape. Digital documents can be spoken by speech synthesizers. With the advent of optical scanning, improved speech-recognition software, and the widespread creation of electronic documents, computer-based speech has become a realistic option for both vocational and pleasure reading. This technology has reduced the dependence of visually disabled readers on sighted assistants. It should be noted, however, that the sequential nature of auditory displays makes nonsequential text reading difficult. Some screen reader programs (such as JAWS by Freedom Scientific) have included special functions to help with nonsequential reading (e.g., a function that groups all hyperlinks on a web page into a single column).
Between 15,000 and 85,000 Americans use Braille (Legge et al., 1999). Although Braille reading speeds average about a factor of two lower than print reading speeds, Braille is especially valuable in contexts in which magnifiers or auditory displays are inconvenient. Word processing documents can be converted to Braille codes by software and embossed on Braille printers. Braille note takers (for example, Braille ‘n Speak by Freedom Scientific or BrailleNote by PulseData) permit users to type Braille on a compact keyboard for later reproduction by synthetic speech, print, or embossed Braille.
There is a debate over the mode of instruction for teaching visually impaired children to read—print, Braille, or tape. In one view, children with low vision should learn to read print because the majority of written material appears in print. In another view, children should learn to read Braille because it is often more
convenient than reading with a magnifier, and because it is hard to learn later in life when vision may decrease further. Almost everyone agrees that strict reliance on audio recordings is undesirable because it results in poor spelling and even illiteracy.
Measuring Reading
Reading performance has been evaluated in many ways. A brief review follows. Clinical tests of reading are discussed in the next section.
Reading Acuity and Critical Print Size. Reading acuity refers to the measurement of visual acuity using a test chart containing paragraphs, sentences, or words in typeset print. The test material for reading acuity is more congested and complex than the letter chart, which has relatively widely separated letters. Not only is the reading material more crowded, but there is more spatial integration required to correctly recognize individual words and word strings. Reading acuity is highly correlated with letter acuity, although some people with low vision, particularly those with macular degeneration, have poorer reading acuity than letter chart acuity (Lovie-Kitchin & Bailey, 1981). Reading acuity is typically measured at a near viewing distance such as 40 cm.
Newsprint, held at a distance of 40 cm (16 inches), would typically be at the limit of resolution of someone with reading acuity of 20/50. Accordingly, someone with a reading acuity poorer than 20/60 (a common definition of low vision) would be unable to read newsprint without bringing the page closer to the eye or using some other form of magnification.
A newer concept is critical print size. While reading acuity documents the angular size of the smallest print size for which reading is possible, a somewhat larger print size is required for fluent, effective reading. For a given viewing distance, the critical print size, two or more times larger than acuity letters, is the print size beyond which the size of characters no longer inhibits reading performance. Although optical and closed circuit television magnifiers are often prescribed to magnify selected specimens of printed material to achieve the observer’s critical
print size, it is only recently that critical print size has received attention from researchers.
For example, consider a person with 20/200 reading acuity whose critical print size is 3 times larger than their acuity limit. If 4-fold magnification brings typical newsprint to this person’s acuity limit, then 12-fold magnification would be necessary to reach the critical print size for effective reading. Although 4-fold magnification might be easily accomplished with a hand-held optical magnifier, 12-fold magnification could require a more specialized magnifying device and a smaller field of view. The important point is that the nature of the prescribed magnifier, as well as the functional outcome, may depend on whether text letters are enlarged to the acuity limit, to the critical print size, or to an intermediate size.
DeMarco & Massof (1997) have surveyed the distribution of print sizes in 10 different sections of 100 U.S. newspapers. Median print sizes range from M = 0.78 (stock listings) to M = 1.21 (comic strips), corresponding to Snellen sizes (at 40 cm) of 20/40 to 20/60. (M-units indicate the distance in meters at which the letter height subtends 5 minutes of arc. For example, 1.0 M print subtends 5 minutes at a distance of 1 meter and is 1.45 mm high.) Anyone with a critical print size larger than 20/60 would be at a disadvantage in reading text similar to newspapers at a distance of 40 cm. At that distance, they would either read slowly (if at all) or would require some magnification.
Reading acuity and critical print size are familiar measures to eye care professionals because of their similarity to letter acuity. The following measures are less familiar and have been used more in rehabilitation or research contexts.
Reading Speed. Reading speed, in words per minute (wpm), has been widely used in psychophysical studies because it can be measured objectively, is reproducible, and is sensitive to variations in visual parameters (Carver, 1990; Legge, Pelli, et al., 1985; Tinker, 1963). Reading speed is a measure that reflects the dynamic nature of reading. One problem with this measure is that reading speed depends on the difficulty level of the reading material. Factors such as the component words, the sentence structure, and the simplicity or complexity of the content necessarily cause variations in difficulty
from one passage to the next or from one observer to another. Mean word length varies from passage to passage, increasing with text difficulty. Carver (1990) has shown that differences in speed due to text difficulty can be reduced by measuring reading speed in “standard-length words” per minute, wherein each six characters count as one standard-length word. Carver has shown that, on average, a subject’s reading speed is about constant in standard-length wpm across text difficulty, provided the grade level of the text does not exceed the reading level of the subject.
Average prose reading speed in English for normally sighted adults is about 250 wpm (cf. Legge et al., 1999). Whittaker & Lovie-Kitchin (1993) have identified three slower rates, based on clinical experience, associated with different levels of function in people with low vision: (1) spot reading (44 wpm), adequate for many tasks of daily life, such as reading mail, recipes, and labels; (2) fluent reading (88 wpm), and (3) high fluent reading (176 wpm). Note that their high fluent rate is well below the typically cited mean value for normally sighted reading speed. A sustainable reading speed of 176 wpm, while still slow for a normally sighted reader, would probably be adequate for meeting the needs of all but the most reading-intensive jobs. However, a person whose maximum reading speed is 90 wpm or less is functionally disadvantaged in reading, lying more than two standard deviations below the normal mean (Legge et al., 1992).
Accuracy. The percentage of words read aloud correctly is sometimes used, especially in cases in which it is expected that faulty control of eye movement may lead to missed portions of text or if the person is inclined to guess.
Endurance. Sometimes people with low vision can read rapidly for a short period, but the motor demands of a magnifier or the nature of the eye condition precludes lengthy sustained reading. There has been little study of reading endurance, but it is surely an important issue for some types of work.
Comprehension. Comprehension is tested in many cognitive and educational studies of reading. Standardized tests, such as the SAT or GRE, evaluate comprehension. It appears that comprehension is not much affected by eye condition or a person’s maximum reading speed
(Legge, Ross, Maxwell, & Luebker, 1989). But there is some evidence that a greater cognitive load is associated with low vision reading. Dickinson & Rabbitt (1991) tested reading comprehension in normally sighted subjects with simulated low vision (optical blur and distortion). Free recall performance was impaired, although performance on multiple-choice questions was not. Further research is required to show whether these subtle effects are found in low vision reading.
Eye Movements. Saccade lengths and fixation times have been widely measured in linguistic and cognitive studies of reading. It is technically difficult to measure eye movements in people with low vision, especially those with scotomas (blind spots) in central vision. This is because eye-tracking hardware usually relies on a subject’s careful fixation for proper calibration. Nevertheless, there have been several studies of eye movements in people with central scotomas from macular disease (Bullimore & Bailey, 1995; Rumney & Leat, 1994; Trauzettel-Klosinski et al., 1994). These studies typically find that slower reading is primarily due to abnormally short saccades, while fixation times are more nearly normal. Legge et al. (1997) have provided a theoretical analysis and computer simulation of saccade behavior in the presence of central scotomas.
Questionnaires. Surveys of visual status usually include questions about reading behavior, and standardized measuring instruments, such as the National Eye Institute Visual Function Questionnaire (discussed below), also include questions about reading. An important issue is the correlation between people’s responses to these questions and their actual reading performance. In one study, a discrepancy was found in the responses of some people with mild visual impairment between measured reading speed and self-reports of reading difficulty on the Activities of Daily Vision questionnaire (Friedman et al., 1999). A small proportion of subjects reported minimal difficulty in reading newsprint despite measured reading speeds less than 80 wpm. The authors suggested that the discrepancy occurred for people who were undergoing changes in visual status during an acute phase of eye disease; they were not yet fully aware of the decline in their reading ability.
Clinical Tests of Reading
There are numerous clinical tests of reading acuity. Some use continuous text (e.g., Sloan & Brown, 1963), and some use randomly ordered words (Bailey & Lovie, 1980). A few reading charts have been designed to standardize the length and difficulty of the reading task in order to facilitate the quantitative assessment of reading speed in either clinical or research environments. These include the charts described in Bailey & Lovie (1980); the MNREAD chart (Mansfield et al., 1996); and the Colenbrander Chart (Precision Vision Catalogue). Figure 3-1 shows sample MNREAD data for a normally sighted subject and a person with age-related macular degeneration.
The Pepper Visual Skills for Reading Test (Baldasare et al., 1986) was designed to measure the effects of word length, line spacing, and other attributes of text on reading by visually impaired people.
Clinical tests of reading performance have been found to correlate with real-world reading performance in the home (West, Rubin, et al., 1997) or with magnifier-aided reading (Ahn & Legge, 1995; Lovie-Kitchin et al., 2000). They are better predictors of real-world reading performance than such standard clinical tests as letter acuity.
Design Characteristics for a Clinical Reading Test
There is consensus that tests of reading vision should provide not only a measure of the smallest print (angular size) that can be read, but also an assessment of reading speed as a function of print size. Reading acuity, critical print size, and maximum reading speed are three important parameters for characterizing reading vision. Some reading acuity tests are composed of unrelated words, but, for testing functional reading ability, tests with continuous text are preferred because the task is more representative of real-world reading tasks. The results from tests composed of unrelated words and continuous text are often similar; reading speeds are faster for continuous text than unrelated words, but the speeds are highly correlated across subjects (Legge, Ross, Luebker, & LaMay, 1989).
![](/openbook/0309083486/xhtml/images/p20005792g139001.jpg)
FIGURE 3-1. Illustrative data from the MNREAD Reading Acuity Chart for a subject with normal vision (upper curve) and a subject with age-related macular degeneration (lower curve). Reading speeds were obtained by measuring the time to read sentences composed of 60 characters (10 words) at each of several print sizes. Three parameters are shown for each curve: reading acuity (arrow at the left end), critical print size (arrow at the bend), and reading speed (dashed line). Source: Gordon E. Legge, unpublished.
For reading tests with continuous text, the charts should follow the same design principles that are recommended for letter charts for measuring distance visual acuity. The task should be essentially the same at each of the size levels. This requires a logarithmic progression of size, standardization of the typeface, and standardization of the layout and reading difficulty throughout the chart. The length of rows and the spacings between letters, between rows, and between
adjacent size levels should be kept proportional to print size. The words and composition of the text should be such that the different passages have approximately equal difficulty. The number of characters per row should be approximately the same throughout the chart.
In order to ensure that reading acuity can be measured for all people, the range of print size should extend to print so small that it is beyond the resolution limit of people with normal vision when viewing the chart from a standard reading distance (e.g., 40 cm). The largest print size should be as great as practical, to accommodate the widest range of low vision subjects. High-quality printing is required to achieve good rendition of the print at the very small sizes.
The text used in reading tests should be simple enough to be read easily by most persons with the expected level of literacy. Simpler text may be necessary for testing young children. The choice of the length of the passages of text requires compromises between (a) sufficient text to estimate reading performance, (b) too much text to fit on the chart or screen at the large print end, and (c) time required by low vision subjects to read through the passages. The text may be single sentences or sequences of sentences forming paragraphs. From one print size to the next, there should not be a continuity of the story line. The average number of character spaces per row should be kept approximately the same at all size levels. If reading speeds are to be quantified, the length of the text samples at each size level should be specified in terms of number of words or characters. It is desirable that there be different versions of the chart available, should the test need to be repeated. The typeface should be representative of commonly used type. Most reading charts use Times Roman or a similar typeface.
Print size should be expressed in angular terms when measuring reading acuity and determining profiles of reading speed as a function of print size. Specifying both the height of the print and viewing distance provides a measure of angle. The height of letters in typeset print may be characterized by the height of the lowercase letters that have neither ascending nor descending limbs; the lowercase x is representative of such letters. The “x-height” may be expressed in millimeters or inches, but it is most common for M-units or points to
be used to specify print size in reading tests. M-units indicate the distance in meters at which the letter height subtends 5 minutes of arc. For example, 1.0 M print at a distance of 1 meter subtends 5 minutes, is 1.45 mm high, and is equivalent to a 20/20 Snellen letter. At a standard reading distance of 40 cm, 1 M letters have the angular size of 20/50 Snellen letters. Points are units traditionally used in the printing industry: 1 point = 1/72 inch. For a sample of typeset print, the size in points measures from the top of the ascenders to the bottom of the descenders. For font styles such as those most commonly used in newspapers and books (e.g. Times, Century, Schoolbook), the x-height is approximately one-half of the total height of the print sample. Thus, in the Times font, 8 point print has an x-height of 4/72 inch = 1.41 mm. If there is a fixed reading distance, the print size may be labeled in terms of angular size as Snellen fractions or logMAR values.
Standardizing Testing Procedures
The conventional or traditional reading distance is 40 cm, although other viewing distances may be used. Persons with impaired vision may need or may prefer to hold the reading material much closer. Logarithmic print scaling can facilitate adjusting scores to allow for the change in distance. For any testing of reading vision, it is important to ensure that the person being tested has the accommodation and/or appropriate correction to ensure that the eye is in good focus for the viewing distance that is being used. In some occupational tasks, there may be constraints on the range of reading distances used in the workplace. For tests of reading vision to have functional relevance, it is appropriate to place the same constraints on viewing distances during the reading test.
When testing the ability to read, it is important that an appropriate optical correction be worn to ensure an in-focus image. People younger than the mid-forties should normally wear their usual distance vision glasses or contact lenses during testing. Their accommodation will normally be sufficient to focus clearly on the printed material at common reading distances.
Older persons with limited or no accommodation ability are likely to require reading glasses in order to achieve clear vision when performing reading tasks at close distances. The reading glasses may be in the form of single vision reading lenses, bifocal, trifocal, or progressive addition lenses. Some older persons with myopia may remove their distance vision glasses to achieve clear vision at a close distance. If a test of reading vision is being conducted at a set reading distance, care should be taken to ensure that any reading glasses provide the best focus for the test distance.
Real-world reading is typically binocular, so it is appropriate to test reading vision binocularly. (For specialized applications, it may be desirable to test reading performance in the left and right eyes separately.) Uniform instructions should apply across subjects and across print sizes for a given subject. If reading speed is measured, the instructions should promote the same reading strategy across all conditions, e.g., “Read the passage aloud, reading as quickly and accurately as possible.” Most tests of reading vision use reading aloud because it is easier to score objectively. When reading aloud at faster speeds, the task of vocalization may limit performance. In principle, silent reading speeds can be measured by monitoring eye movements or by having the subject indicate when the text sample has been completed, although these techniques may be less reliable for purposes of disability determination.
Charts should be uniformly illuminated in a manner that avoids unwanted reflection glare. For compatibility with conditions for testing distance visual acuity, the luminance of the white background should be in the range 80 to 320 cd/m2 .
Reading acuity is given by indicating the smallest print that can be read and specifying the test distance. Reading errors may be counted and recorded. There should be rules for what constitutes successful reading of a text sample. When reading speeds are to be quantified, there should be rules about extracting the reading speeds and for determining the smallest print.
All of the above criteria assume that the subject is literate in the language in which reading is being tested. Our discussion is based on
testing in English, but reading tests are available or under development in other languages. We have not evaluated these here.
Evaluation
To evaluate reading as a potential measure of functional vision, the committee reviewed what is known about (1) the relationship to standard clinical measures of visual impairment and (2) common stimulus variables that may affect reading performance. Before discussing these topics in detail, we address some up-front questions.
What Is the Cause of the Reading Disability?
While most types of impaired vision result in reading problems, it is not the case that all reading problems result from impaired vision. In the United States, low literacy is a major societal problem. According to the National Institute for Literacy (http://novel.nifl.gov/nifl/faqs.html), the 1992 National Adult Literacy Survey revealed that between 21 and 23 percent of the adult population had level 1 literacy skills (i.e., they were unable to fill out most forms or read a simple story to a child). Other nonvisual causes of poor reading include testing in a nonnative language, dyslexia (or related higher-level disorders), or cognitive dysfunction. It is a policy decision whether it is important to ascertain the visual origins of a reading disability.
Assuming that reading disability needs to be ascribed to visual impairment, a diagnosed eye disorder or measurable visual impairment provides a prima facie case. If a nonvisual cause of reading disability is suspected, documentary evidence should be included in the assessment (e.g., a low score on the Mini-Mental Status Exam would be indicative of cognitive dysfunction). For many purposes of disability determination, rehabilitation specialists conduct an assessment including compulsory reports from eye care professionals to establish the visual origin of the disability. If a standardized measurement of reading performance was a part of disability
determination, it is reasonable to expect that the origin of poor reading (visual or nonvisual) could be established with high confidence in most cases.
Age is a nonvisual factor requiring special consideration. There is evidence that age per se has greater impact on reading speed for people with low vision than for normal readers (Legge et al., 1992). It is also noteworthy that age has a greater impact on more challenging tests of visual function—such as low-contrast, low-luminance acuity—than it does on standard visual acuity (Haegerstrom-Portnoy et al., 1999).
Is There a Scale for Reading Disability?
The measures of reading performance that have received most attention from clinicians and vision researchers are reading acuity, critical print size, and reading speed. These measures are all reactive to eye problems and are to some degree decoupled from one another. How should they be combined in the determination of disability? This problem, like visual field testing, requires a method for collapsing multiple measures into a single score. In principle, there is no greater barrier in constructing a scale for reading disability than for constructing scales for functional acuity or functional field measurements.
Are Reading Measurements Too Variable?
Eye clinicians have had vastly more experience with acuity and field measurements than with reading measurements (although informal use of reading charts is common in refraction). Intuitively, it seems likely that acuity measurements are tighter and more reliable than reading measurements.
Surprisingly, well-controlled studies of the distribution of normal acuity and the reliability of repeat measurements are few and far between. In part, this is because suitably designed acuity charts and scoring methods have been developed only in recent years (Bailey &
Lovie, 1976; Ferris et al., 1982). Leat et al. (1999) summarized the small number of tightly controlled studies in which all subjects were screened to have no eye disorders, best refraction was used, and acuity charts were used with sufficient lines to avoid floor effects. Across studies (see Table 1 of Leat et al.), mean normal acuity was about −0.10 logMAR (Snellen 20/16) with a standard deviation of about 0.1 log unit (i.e., percentage error of about 26 percent). An identical standard deviation of 0.1 log units was obtained for a normal control group in a study of reading speeds for people with low vision by Legge et al. (1992). In a large sample study of older subjects in Marin County, California described by Haegerstrom-Portnoy et al. (1999), data from 748 subjects were analyzed2 to compare the standard deviations of acuity measurement (near acuity measured with the SKILL test) and reading speed measurement (from the Pepper test). For subjects less than 65 years of age, the standard deviations were 0.09 and 0.10 log units for acuity and reading speed, respectively. The standard deviations of both tests grew with increasing age. For subjects over 90 years of age, the values were 0.18 and 0.24 log units for acuity and reading, respectively.
Although more studies are needed of both acuity and parameters of reading, preferably stratified by age, existing evidence indicates that variability of normal reading performance is not much greater than variability in acuity values.
Should Magnifiers Be Allowed During Testing?
Rehabilitation includes the prescription of reading magnifiers. Magnifiers increase the range of accessible print sizes and reduce the impact of vision impairment on reading. In other words, magnifiers reduce reading disability, although they do not affect visual impairment. It is reasonable to argue, therefore, that a person’s reading disability should be assessed while using a magnifier. This would be analogous to evaluating the mobility of a paraplegic person in a wheelchair.
In opposition, it may be argued that the magnifier introduces too many uncertainties into the assessment of reading disability. For instance: Is it practical to use a suitable magnifier for reading tasks in the person’s line of work? Has an appropriate magnifier been prescribed? Will the trade-off of field and print size, inherent in magnifiers, adversely affect work performance?
A hybrid approach might be to measure reading performance with and without a prescribed magnifier and compute reading disability as a weighted combination.
What About Manipulation?
When financial benefits are on the line, it is desirable to use tests that are immune to the subject’s manipulation. Any behavioral tests of best sensory, motor, or cognitive function are vulnerable in this sense. Certainly, poor reading could be faked by a subject. As in the case of acuity testing, however, an experienced examiner could use subtleties of the test to ferret out manipulation. For instance, do the critical print size and reading acuity change appropriately when the viewing distance is halved or doubled? Is the person’s reading acuity close to their distance Snellen acuity? How easily did they read the printed instructions given to them before the test? In short, assessment of reading disability may be no more vulnerable to manipulation than tests of visual impairment, such as acuity.
Relationship of Reading Performance to Impairments of Visual Functions
Reduced acuity, reduced contrast sensitivity, and loss of macular function are the primary visual impairments affecting reading for people with low vision.
Acuity
Reading the newspaper at a normal reading distance (40 cm or 16 in) is frequently taken as an example of a visually demanding task. According to DeMarco & Massof (1997), 75 percent of material in all newspapers exceeds 0.8 M print size. Letters of this size would be at the acuity limit of a person with 20/40 acuity (the print on medicine bottle labels can often be as small as 0.4 M).
Typically, people need letters larger than their acuity limit to read quickly and without fatigue. The increased print size for fluent reading—the acuity reserve or critical print size—is a factor of 2 or more larger than acuity letters (Mansfield et al., 1996; Whittaker & Lovie-Kitchin, 1993). Accordingly, even a person with 20/40 acuity would be at a visual disadvantage in a job with demands equivalent to newspaper reading.
Clinical tests of letter or reading acuity determine the tiniest letters that can be read. They are reasonably good predictors of the range of legible print sizes under optimal conditions. Conventional letter acuity measurements are not, however, good predictors of reading speed when adequate magnification compensates for acuity limitations (Legge et al., 1992; Whittaker & Lovie-Kitchin, 1993). A recent report indicates that near acuity measurement, based on text or unrelated words, is predictive of reading speeds in people with macular degeneration (Lovie-Kitchin et al., 2000).
Fields
Broadly speaking, field loss can be divided into macular loss, peripheral loss, or hemianopsia. Blind spots (scotomas) in the macular region at the center of the visual field are common in macular degeneration, the leading cause of low vision in the United States. When macular scotomas are present, reading is almost always severely affected (cf. Faye, 1984). The tight link between macular loss and reading difficulty is a strong argument for considering the status of the central fields in disability determination.
Why is macular field loss so detrimental to reading? People with central scotomas typically adopt a region of peripheral vision for fixation termed the preferred retinal locus. If peripheral vision were simply a low-resolution version of central vision, reading difficulty could be remedied by magnifying text presented to the preferred retinal locus. Magnification does help, but it does not restore people with macular degeneration to normal reading speed. Nor is it a question of eye movement control. Even when a text presentation method is used that minimizes the need for eye movements, people with central scotomas still read slowly (Rubin & Turano, 1994). Recently, Legge et al. (2001) reported evidence that the number of characters recognized in parallel, termed the visual span, shrinks in peripheral vision and imposes an inescapable bottleneck on reading speed.
Peripheral loss can affect reading if the region of remaining central vision gets so small that only a few letters can be seen at a time. This form of “tunnel vision” can occur in advanced cases of glaucoma or retinitis pigmentosa. In such cases, text composed of very large letters can be more difficult to read than text containing smaller letters. A similar situation occurs in some cases of macular scotoma in which the patient retains a small region of foveal function (Fletcher et al., 1999).
Hemianopsia refers to complete loss of either the left or right side of the visual field, usually due to stroke. Although both types of hemianopsias reduce reading speed, loss of the right visual field tends to produce greater deficits than loss in the left visual field (Trauzettel-Klosinski & Brendler, 1998).
Contrast Sensitivity
Contrast sensitivity deficits can be present when acuity and field are relatively intact (Elliott & Whitaker, 1992b), a finding that has motivated the development of special charts for measuring contrast sensitivity discussed in Chapter 2 (cf. Pelli et al., 1988).
Deficits in contrast sensitivity are related to reduced reading performance (Leat & Woo, 1997; Rubin & Legge, 1989; Whittaker & Lovie-Kitchin, 1993). The latter authors found that Pelli-Robson
contrast sensitivity less than 1.05 was invariably associated with reduced reading speed. Rubin & Legge (1989) proposed that a subset of people with low vision (those with cataract and other forms of cloudy ocular media) shows normal reading performance after loss of contrast sensitivity is accounted for. For these people, contrast sensitivity can be used to predict reading performance.
The contrast polarity of text (white on black or black on white letters) has little effect on normal reading, but it can affect low vision reading. People with cloudy ocular media suffer from glare due to light scatter within the eye. They usually read white on black text faster and with better acuity than conventional black on white text (Legge, Rubin, & Schleske, 1987). Electronic magnifiers usually include a contrast polarity switch.
Visual Functions That Have Little Effect on Reading
For people with normal binocular vision, in which the two eyes are nearly matched, there are only slight effects on acuity or reading performance of binocular versus monocular viewing (Jones & Lee, 1981; Legge, Pelli, et al., 1985; Sheedy et al., 1986). For instance, Sheedy et al. found only a 3.7 percent reading speed advantage for binocular viewing. Stereo depth is not considered to be relevant to reading.
People with impaired vision usually see better with one eye than the other. When the interocular difference in vision is large, the better eye undoubtedly governs reading performance. There are clinical reports that the poorer eye sometimes interferes with reading. There is need for research to better understand binocular interactions in low vision.
Eye disease frequently results in color deficits, sometimes combining with inherited colored defects. The acquired color defects vary widely in type and severity. Although bright colors can provide helpful cues for people with low vision, there is no evidence that color coding facilitates reading. Based on current knowledge, maximizing luminance contrast is the best way to enhance text legibility, with color contrast secondary.
Search capacity is probably relevant to nonsequential reading (e.g., searching for hyperlinks on a web page.) The relevance of findings on preattentive and serial search to skimming and searching in reading is not well established.
Stimulus Properties and the “Reading Envelope”
It is important to understand how reading performance depends on the stimulus properties of the text for two reasons. First, reading as a real-world task is subject to wide variations in viewing conditions and text characteristics (undoubtedly true for on-the-job applications). We need to understand how these stimulus factors affect performance. Second, in designing tests of reading vision, it is important to know how viewing conditions affect performance.
Beginning with the seminal studies of Tinker and colleagues (summarized in Tinker, 1963), there have been many studies of the effects of text characteristics on reading performance in normal vision. Many of these studies help to define the reserve or “envelope” for reading—that is, the range of conditions over which fluent reading remains possible. Important examples include the effects of print size, text contrast and text luminance.
Reading performance is limited by angular print size, which is jointly determined by physical print size on the page (or screen) and viewing distance. Normal reading speed is at maximum for angular print sizes over a 10-fold range from about 0.2 to 2.0 deg (Legge, Pelli, et al., 1985). For people with reduced acuity, this range is contracted.
Normal reading speed is roughly independent of text contrast from 100 percent contrast down to 10 percent or a little lower, there being some interaction with character size (Legge, Rubin, & Luebker, 1987). It is this tolerance to contrast reduction that enables normally sighted people to read low-contrast xerox copies or poor-quality computer displays. It is not known if reduced contrast affects reading endurance. People with reduced contrast sensitivity have less tolerance to loss of text contrast.
Legge & Rubin (1986) showed that normally sighted reading speeds decrease only slightly over a wide range of moderate to low photopic text luminance, but they decrease more rapidly in scotopic vision.
Sloan (1969) measured acuity versus luminance for a diverse group of subjects with low vision. While some reached maximum acuity at lower luminance levels than required by normally sighted subjects, some subjects with macular degeneration required higher luminance to achieve maximum acuity. Sloan subsequently recommended bright task lighting as a reading aid for people with macular degeneration (Sloan et al., 1973). Bullimore and Bailey (1995) studied reading eye movements in persons with macular degeneration and found substantial changes in reading speeds with changes in illumination. In a recent study, Aquilante et al. (2000) measured reading speeds for normal subjects and subjects with central field loss from macular degeneration. They varied both luminance and print size. They found no qualitative difference between the normal and central loss subjects; both showed decreased reading speed when the luminance was low, with the effect of luminance increasing for print sizes near an individual’s acuity limit. More research is needed to clarify the effects of luminance on reading in people with low vision.
The reading envelope of normal vision encompasses a wide range of print sizes, contrasts, and luminance levels. As a result, people can read fluently over a wide range of environmental conditions— variation in viewing distance, daylight or twilight, etc. One important effect of visual impairment is to shrink the envelope. Some people with mild visual impairment may read normally under optimal environmental conditions (e.g., the conditions in effect for a clinical test of reading), but they show a sharp decline with modest changes in text contrast or light level.
Recommendations
Reading is a necessary component of many jobs in the modern economy. Anyone with a reading disability due to vision impairment is restricted in the range of jobs available to them, and faces
impediments in many other jobs. Almost everyone with low vision encounters reading difficulty. We can be confident that most people meeting the SSA’s medical listing criteria for visual impairment (recommended by the committee to include tests of acuity, field, and contrast sensitivity) will encounter problems with visual reading. For this reason, an additional test of reading performance is unnecessary for those who meet the medical listing criteria.
But some people with visual impairments who fail to meet the listing criteria may experience reading problems that hamper their employability. For example, people with macular problems, notoriously difficult to measure with contemporary field tests, frequently have severe reading disabilities. Others may experience reading difficulty due to combined effects of acuity, contrast, or field deficits that, by themselves, do not meet the criteria.
The committee recommends that a test of reading vision should be included as a key component in the assessment of individuals with vision impairment who receive vocational assessment when their impairments do not meet the medical listing criteria. This should be implemented as soon as a well-normed reading test can be shown to meet test standards established by SSA. Any such assessment may take into consideration the impact on reading of viewing conditions or circumstances associated with an individual’s vocational niche (e.g., a person whose reading performance deteriorates rapidly at moderate or low luminance may no longer be able to hold a job that requires reading at low light levels).
Three parameters of reading vision should be taken into account in evaluating disability. A person with reading acuity equivalent to 20/60 or worse will be unable to resolve text similar in size to newsprint if it is at a normal working distance of 40 cm (16 inches). A person whose critical print size is equivalent to 20/60 or less will be unable to read fluently most text in newspapers and other documents of equivalent print size if they are held at a normal reading distance. A person with a maximum reading speed of 90 words per minute or less will be functionally disadvantaged in reading.
Psychophysical and clinical studies of reading and vision have reached the point at which appropriate tests exist or can be designed to measure
reading disability. This section has outlined the findings and design principles against which to judge candidate tests. We recommend the following criteria for tests of reading vision:
-
Visual characteristics that are consistent with contemporary standards for acuity tests, including a logarithmic progression of print sizes;
-
A range of print sizes containing large enough print to be useful with most visually impaired people and small enough to reliably measure reading acuity in normally sighted people;
-
Text passages equated in layout across print sizes;
-
Reproducible rules for estimating reading acuity, critical print size, and reading speed;
-
Binocular testing, unless one eye interferes with the other in reading and the problem can be addressed by covering the interfering eye; and
-
Text passages representative in font and letter spacing of commonly encountered real-world texts.
We recommend additional research to establish in more detail the distributions of reading acuity, critical print size, and reading speed in different age groups and the relationships between these measures and performance of work-related activities and the important tasks of daily life.
ORIENTATION AND MOBILITY
Ambulatory Mobility
Independent travel on foot and on public transit—an important prerequisite for employment and independent living—is severely affected by blindness and visual impairment. Indeed, an entire orientation and mobility (O&M) profession has evolved to address this
problem (Blasch et al., 1997). The visual requirements for independent travel have received rather limited study compared with other tasks, such as flying a plane or driving an automobile, but enough is known for some conclusions to be drawn.
As the O&M acronym suggests, the independent travel problem for blind and visually impaired people is commonly broken down into two parts. Mobility is commonly thought of as the problem of maintaining a safe and straight path through the environment, avoiding obstacles, collisions, dropoffs, and excessive veering. Orientation is the more global navigation or wayfinding aspect of the problem, involving finding one’s way from A to B and maintaining a knowledge of where one is, what direction one is facing, etc. (Blasch et al., 1997).
Theories of Orientation and Mobility
The use of visual information for maintaining a path through the environment and steering around obstacles has long been studied in terms of optical flow (Gibson, 1958, 1979). Navigation, a higher-order task, involves more elaborate spatial representations (Strelow, 1985). For example, “cognitive maps” (Tolman, 1948) loosely refer to a viewpoint-independent representation of spatial layout in environmental (allocentric) coordinates.
Evidence for the salience of landmarks in spatial representations has been used to argue for navigation based on “route memory” rather than cognitive maps. Route memory consists of chains of associations between perceived landmarks and motor movements (Schoelkopf & Mallot, 1995). Route memory relies on egocentric coordinates, and is ineffective for judging the relations between locations not on the same route. Route memory has often been ascribed to people with blindness or low vision to explain deficiencies in spatial tasks (cf. Thinus-Blanc & Gaunet, 1997). Path integration (Loomis et al., 1999) is yet another model for the cognitive aspect of navigation, probably most useful in exploring unknown environments.
It is likely that travelers invoke cognitive maps, route memory, or path integration depending on their task demands or training. The studies
suggest the importance of visual landmarks in forming all of these cognitive representations, and impaired vision may make them harder to learn because of reduced access to landmarks.
Travel Needs of Blind and Partially Sighted Individuals
As mentioned earlier, the profession of orientation and mobility instruction has evolved to address the rehabilitation needs of blind and visually impaired persons relating to travel. A wide variety of training techniques and assistive devices have been developed specifically for this problem. To address mobility, people with little or no functional vision have to rely on canes or guide dogs in combination with training in the use of auditory and tactile cues, such as traffic sounds, echolocation, and surface texture, and various other techniques for safe travel. The cane allows direct detection of obstacles, surfaces, dropoffs and shorelines (walls, edges of pathways, etc., which have to be paralleled during travel) within the zone covered by its scanning pattern. It also allows indirect detection of large objects (walls, building entrances, etc.) via echoes derived from the tapping of its tip on the ground (Wiener & Lawson, 1997). Traffic sounds are used to orient one’s path relative to the street and determine when crossings can be safely made. In the last resort, drivers may be able to see the white cane and avoid the blind pedestrian.
Travelers with significant residual vision can supplement these methods with visual detection of large obstacles, shorelines, and oncoming traffic. However, detection of other hazards such curbs, dropoffs, small objects on the ground that could be tripped over, and intersection crossing information, depend on the nature and amount of an individual’s residual vision. Important factors include the effectiveness of the individual’s vision function in the real world of varying light levels and viewing conditions, as well as the optical aids and training he receives to optimize its use (Geruschat & Smith, 1997).
The orientation aspect of travel is considerably complicated when vision is impaired or absent. This applies particularly to travel in unfamiliar surroundings—which many blind and visually impaired
people consequently avoid. Access to landmarks and printed navigational signs, which are the sighted traveler’s main cues, is effectively eliminated or severely reduced (Arditi & Brabyn, 2000). Blind travelers are taught a number of orientation techniques that rely heavily on memory, as well as cues involving sounds, smell, and touch. Examples include feeling the direction of the sun shining on the face, remembering the sounds and smells of different types of shops, memorizing the number of blocks along a route, and remembering as many cues as possible along a route that has been travelled before (Long & Hill, 1997). Persons with some residual vision can add visual information about large objects that can serve as landmarks in the environment. Signs, once located, may be read with a hand-held telescope if visual function is sufficient.
Direct Measures of Orientation and Mobility Performance
Beginning in the 1970s, researchers have used various methods of quantifying ambulatory travel performance for studies on electronic travel aids for blind persons and on the visual factors contributing to travel performance deficits. Such measurements are difficult partly because of the number of variables encountered (traffic, weather, presence of obstacles, etc.) in outdoor travel routes. Early efforts (e.g., Armstrong, 1972, 1975) used an outdoor test route to measure various aspects of safety, efficiency, and stress in an effort to evaluate new devices. Brabyn and Strelow (1977) designed a position sensing system using a computer for objectively measuring locomotion of blind subjects indoors. Dodds et al. (1983) extended the Armstrong technique with video recordings and the use of numerous parameters, such as time taken, productive walking index (percentage of time spent walking), cane and body contacts with obstacles and shorelines, steps taken, and curb incidents and lateral position on the pavement. Later researchers followed this general approach, using a wide variety of indoor and outdoor courses of varying realism and complexity (e.g., Haymes et al., 1996; Kuyk et al., 1998; Long et al., 1990; Lovie-Kitchin et al., 1990; Marron & Bailey, 1982). Some researchers lumped many of the items together as “mobility incidents” (Geruschat et al., 1998). Variations included the use of a secondary task (reaction time to
randomly emitted tones) to gauge the mental effort imposed by the mobility task (Turano et al., 1998). Most of these approaches dealt only with the mobility aspect of the travel problem; fewer efforts have been made to measure the orientation aspect of travel skills. Crandall et al. (1999) used a variety of indoor and outdoor travel routes to test blind subjects’ ability to find particular destinations with and without an infrared navigation device (Talking Signs), with the percentage of successful route completions as the main measure.
Due to difficulties such as specifying and implementing standard travel routes in different localities, defining measures that would be common to all routes, and incorporating orientation as well as mobility measurement, no standardized measure of orientation and mobility performance has been widely adopted. The use of simulators has received little attention in this field and could be a possible avenue for standardization. At the present time, however, it would be difficult to recommend any particular test as a benchmark for determining disability in terms of travel performance.
Aspects of Vision Function Affecting Orientation and Mobility
In controlled environments, in the absence of such hazards as dropoffs, small low-lying objects to trip over, or fast approaching traffic, the mobility aspects of travel can be performed with relatively poor vision. Pelli (1987), working with normally sighted subjects with simulated low vision in a controlled environment (a shopping mall), showed that obstacle avoidance is possible with severely reduced vision (acuity reduction achieved by blur, contrast reduction, and field restriction.) Studies of obstacle avoidance with low vision subjects (again, mainly in well-controlled environments) have usually shown that acuity level is not important, contrast sensitivity is somewhat important, and the total extent of the visual field is of major importance (Haymes et al., 1996; Kuyk et al., 1998; Long et al., 1990; Lovie-Kitchin et al., 1990; Marron & Bailey, 1982).
In the complex, uncontrolled environments found in the real world beyond experimental studies, hazards abound that are not easy for the
individual with impaired vision to detect. Also, visual function is greatly reduced under the less than ideal conditions found in the real world (Brabyn et al., 2000; Kuyk et al., 1996). A brief discussion follows of the way the various hazards and problems interact with different measures of visual function.
Acuity. Laboratory studies aside, acuity is clearly a factor in real-world orientation and mobility. It is necessary, for example, in detecting and avoiding small objects or surface irregularities on the ground in front of the traveler in order to avoid tripping. Even making visual determinations about surface texture (e.g., is the patch ahead water or ice) requires acuity. Inability to see fine detail may make it difficult to distinguish between a dropoff and a shadow (Guth & Rieser, 1997). Some degree of acuity is obviously important in recognition of environmental landmarks. However the major area in which acuity is vital is in finding and reading the signs on which one depends for orientation and navigation. Signs are designed in overall size and in print size to be located and read by individuals with normal or near-normal acuity from the distance at which their particular information is needed (20/40 is a common standard for road signs). A person with, say, 20/100 acuity has to be 2.5 to 5 times as close as the signmaker envisaged in order to find and read the information. This decreases the chance that the individual will be able to find and use the information as intended. (Even if 5-fold magnification via a telescope is used, the viewable area will necessarily be reduced by at least a factor of 25, making location of the sign more difficult.)
Contrast Sensitivity. Visually guided travel is dependent on the ability to see objects (whether large or small) of widely varying contrast against their backgrounds. It is therefore not surprising that a number of studies have found an association between contrast sensitivity and mobility (Geruschat et al., 1998; Kuyk & Elliott, 1999; Kuyk et al., 1998; Marron & Bailey, 1982; Rubin et al., 1994; Turano et al., 1999). In nearly all cases, contrast sensitivity was a far better predictor of mobility performance than acuity (and often the only predictor). Many tripping hazards are of very low contrast. Examples include curbs, step-ups and step-downs, stairways, and dropoffs. Rubin et al. (2001) found an association between contrast sensitivity and self-reported difficulty going down steps. There are also driveway
indentations crossing the sidewalk, wheelchair ramp borders, and discontinuities in the sidewalk pavement, such as the uplifting of one slab slightly above another by an underlying tree root. Good contrast sensitivity is needed to detect these hazards and avoid tripping and is therefore critical in ensuring safety (Geruschat & Smith, 1997).
Color contrast is also important in mobility. For example, it is generally accepted that yellow markings on stairs, etc., can help make low-contrast edges more visible to those with reduced vision.
Visual Fields. Not surprisingly, a number of mobility studies have found a relation between mobility performance and visual fields (Brown et al., 1986; Geruschat et al., 1998; Kuyk et al., 1998; Lovie-Kitchin et al., 1990; Marron & Bailey, 1982). The hazards referred to above lie in the lower visual field, making this part of the field extremely important for safety while walking (Lovie-Kitchin et al., 1990). Visual field defects can also make detection of hazards in other parts of the visual field difficult or unreliable. For example, hemianopic fields make detection of shorelines, traffic, or other hazards on one side or the other very difficult, with potentially catastrophic results. Very narrow visual fields reduce the chances that hazards of any kind will be detected, reducing the value of optic flow, widely thought to be important in mobility (Gibson, 1958). Visual fields also affect orientation. Rieser et al. (1992) asked subjects with low vision to judge from memory the directions and distances between landmarks in their neighborhoods. They found no effect of acuity level, but people with early onset of narrow visual fields tended to perform more poorly than others. Reduced field size could affect other aspects of navigation by reducing the probability of finding or noticing landmarks and navigational signs.
Adaptation and Glare. Difficulty adapting to poor or changing light levels is widely acknowledged to impact the mobility of many visually impaired persons (Szlyk et al., 1990). This problem is closely related to the visual function of glare recovery (see section on Glare and Light/Dark Adaptation in Chapter 2). Geruschat and Smith (1997, p. 63) assert that “The most frequently reported mobility problem for persons with low vision is lighting, inclusive of glare; light adaptation from outdoors to indoors and vice versa; dim and
night lighting; and frequent changes in lighting.” Kuyk et al. (1996) demonstrated that the ability of visually impaired individuals to avoid obstacles is significantly impaired under low illumination. Turano et al. (Turano et al., 1998) found that four of the six most difficult mobility situations for people with retinitis pigmentosa were related to lighting conditions. Certainly, all the problems mentioned above under acuity and contrast are made much worse under poor or changing light conditions. Both acuity and contrast sensitivity fall off rapidly as light level is reduced, making visually impaired persons even more subject to these hazards in dusk or nighttime conditions.
Glare recovery is also an important factor in real-world orientation and mobility. An inability to adapt rapidly to changing light conditions can be disabling for mobility when going from bright sun to indoors or vice versa. Glare recovery declines more with age than many other aspects of vision function (Brabyn, 1999), so this problem is particularly prevalent in older persons. Although it is not commonly measured in younger individuals, it is definitely an aspect of vision that needs to be taken seriously for its negative effects on mobility and safety. However current lack of standardized testing methods makes its adoption as a stand-alone criterion for disability problematic.
Disability glare, or disruption of vision due to veiling glare, can impede the detection and reading of navigational signs against a bright sky, as well as possibly affecting detection of traffic in some circumstances. People with certain forms of low vision, such as retinitis pigmentosa, have particularly severe problems with glare (Turano et al., 1998).
Binocular Vision. There is some evidence that binocular vision or stereopsis is important in mobility, as poor stereopsis has been associated with hip fractures, which in most cases result from falls, in some studies of older populations (Ivers et al., 2000).
Visuocognitive Factors. As noted in the discussion of theories of orientation and mobility, cognitive factors and memory abilities are important in orientation and navigation. For the mobility aspect of the task, the traveler often has to concentrate on the path ahead but be alert to hazards in the lower and horizontal visual fields. Therefore, recently developed tests of divided attention such as the useful field of
view and the attentional visual field may be relevant, as they are in similar situations in the driving task (Brabyn, 1990; Brabyn et al., 1994; Owsley, 1994; Owsley et al., 1991). However these tests go beyond the realm of pure vision function.
Summary and Recommendations
The most important aspects of visual function for safe and efficient ambulatory orientation and mobility are contrast sensitivity, visual fields, and acuity. The next most important variable is adaptation to low or changing light conditions. Disability glare, binocular vision, and visuocognitive functioning are significant but of lesser importance and do not currently lend themselves well to stand-alone tests of visual disability in relation to independent travel. Research is needed on better quantification of these aspects of vision. Pending such standardization, they can be dealt with only by subjectively evaluating any disabling impact on mobility and other activities for individuals not meeting the SSA medical listing criteria.
Driving Mobility
Driving is a specialized type of mobility. Many jobs involve driving, including those that require workers to operate vehicles that transport goods (e.g., interstate truck drivers, local package delivery drivers) and those that involve the transport of people (e.g., bus drivers, taxi drivers). U.S. employment data from 1999 indicate that over 9.5 million persons in the United States were engaged in transportation and material moving occupations, with about half these jobs consisting of operation of a car, bus, or truck (Bureau of Labor Statistics, U.S. Department of Labor, 2000). There is a great deal of emphasis placed on driver safety in U.S. society. The concern is not just whether one can transport the goods or people to their destination, but whether one can do so in a way that does not endanger oneself, coworkers, passengers, or the public.
Investigations into the driving task typically focus on either performance or safety. Performance is usually operationalized as accuracy or latency of a driving maneuver or control input (e.g., staying in lane, braking to avoid a collision), exhibiting certain behaviors (e.g., using mirrors), or other measures, according to some graded scale. Safety is usually defined in terms of adverse driving events, such as crash involvement (e.g., at-fault crashes, injurious crashes) or moving violations (e.g., speeding, failure to obey traffic control devices). Measures of safety are often expressed statistically, such as a risk ratio or odds ratio in which a subgroup of drivers of special interest is compared with a reference group (e.g., visually impaired drivers compared with drivers who have 20/20 visual acuity or better). Although driving performance should be theoretically linked to driving safety, there is little empirical evidence for this link. This is probably due to the fact that there are numerous operator, vehicle, and environmental factors that influence driving performance and the likelihood of being involved in an adverse event (e.g., crash).
The research literature that examines the impact of vision impairment on driving is huge, with a recent review (Owsley & McGwin, 1999) listing over 200 references from the peer-reviewed literature and government publications. Most of these studies are not based on commercial drivers (i.e., those who drive in performing their jobs) but rather on drivers of personal vehicles. One major difference between the driving demands for drivers of personal vehicles and those of commercial drivers is that commercial drivers have high levels of driving exposure (i.e., miles driven per week, time on the road). Exposure is a key for understanding crash risk, since one’s risk increases with exposure to the road. Exposure is also relevant from a fatigue and task-vigilance standpoint. Another noteworthy feature of the vision impairment and driving literature is that it is primarily based on older drivers. This is because vision impairment is more prevalent in late adulthood, and thus questions about the relationship of vision impairment and driving are more easily addressed among older populations. Finally, it is worthwhile to note that although the focus here is on drivers of ground vehicles, the impact of vision impairment on other types of transportation operators is also worthy of consideration (e.g., airline pilots, maritime pilots, rail engineers).
Below we summarize the vision impairment and driving literature, by type of vision impairment. The reader is also referred to recent reviews of this literature for additional details and commentary (North, 1985; Owsley & McGwin, 1999; Owsley, Stalvey, et al., 2001; Shinar & Schieber, 1991).
Visual Acuity
Visual acuity impairment (worse than 20/40 in the better eye or as measured binocularly) can hamper road sign visibility and also the avoidance of some obstacles in the roadway (Higgins et al., 1998; Wood, 1999). With respect to safety, however, visual acuity impairment, in the range that has been studied in detail (20/60 or better), does not appear to threaten road safety, or does so only weakly (Ball et al., 1993; Davison, 1985; Decina & Staplin, 1993; Gresset & Meyer, 1994; Henderson & Burg, 1974; Hills & Burg, 1977; Hofstetter, 1976; Humphriss, 1987; Ivers et al., 1999; Johansson et al., 1996; Marottoli, Cooney, et al., 1994; Marottoli, Richardson, et al., 1998; McCloskey et al., 1994; Owsley, Ball, Sloane, et al., 1991; Owsley, Ball, McGwin, et al., 1998). Drivers with significant acuity impairment tend not to be on the road, either because of state laws removing their licenses when visual acuity drops below a certain statutory requirement or because of self-restriction of their own driving (i.e., they remove themselves from the road or drastically reduce their exposure). Thus, it is difficult to evaluate crash risk in the population of drivers with severe acuity impairment (worse than acuity in the range of 20/70 to 20/100) because these individuals have low or no exposure.
Visual Fields
Studies simulating serious visual field restriction (binocular) have shown that a 40° radius visual field and smaller can compromise some aspects of driving performance (e.g., road sign identification, obstacle avoidance, reaction time) (Wood & Troutbeck, 1992, 1995; Wood et al., 1993). Simulation studies, while useful, must be cautiously
generalized to actual driving performance by those with real visual field restrictions, since it is likely that the impact of a sudden, simulated field restriction is not identical to that of a naturally occurring restriction from an eye disease or neurological disorder. Those with the naturally occurring disorders may develop compensatory mechanisms (e.g., eye and head movements) over time. In several studies where real-world driving performance was assessed, drivers with actual (not simulated) field loss did not exhibit increased driving problems (Cashell, 1970; Council & Allen, 1974; Marottoli et al., 1998). The impact of binocular field restriction on driving performance appears to be an area in need of clarification. In studies addressing crash risk, drivers with severe binocular visual field loss (i.e., significant loss of peripheral vision in both eyes) appear to have twice the crash risk of those without this deficit (Ball et al., 1993; Johnson & Keltner, 1983).
Contrast Sensitivity
There are fewer studies on contrast sensitivity and driving than on acuity and visual field sensitivity, so conclusions about it must be more tentative. Simulated contrast sensitivity impairment appears to be associated with impaired driving performance (Wood & Troutbeck, 1992; Wood et al., 1993). Actual contrast sensitivity impairment in older drivers is associated with crash involvement in analyses not adjusted for confounding factors (Ball et al., 1993; Marottoli et al., 1998; Owsley, 1994). Severe contrast sensitivity impairment (Pelli-Robson score of 1.25 or less) due to cataract is significantly associated with a history of crash involvement, even if present in only one eye (Owsley, Stalvey, et al., 2001).
Visual Search/Attention
Impairment in visual search skills, including deficits in divided attention and slowed processing speed, are associated with crash involvement (Ball et al., 1993; Barrett et al., 1977; Kahneman et al.,
1973; Mihal & Barrett, 1976; Owsley, McGwin, & Ball, 1998). Even when drivers have good visual sensory function (acuity, peripheral vision), they can exhibit deficits in visual search skills (Ball & Owsley, 1991; Ball et al., 1993). This is a relatively common problem among older adults (Rubin et al., 1999). It appears that these sorts of higher-order, visual-processing skills are better predictors of high-risk drivers than visual sensory measures (Ball et al., 1993; Owsley, McGwin, et al., 1998; Rubin et al., 1999). This may be due to the fact that tests of visual attention rely on a more comprehensive set of visual skills, not just good visual sensory input. Although the data are not plentiful, it also appears that poor visual search skills are associated with poor on-road driving performance and performance in a driving simulator (Cushman, 1996; Duchek et al., 1998; Rizzo et al., 1997).
Monocularity
In most drivers the visual function in the two eyes is highly similar. However for a few individuals the visual capabilities of the eyes are drastically different, because of either an ocular or a neurological condition or trauma. There have been a few studies over the years that have examined the role of monocularity in driver safety and performance. In these studies, monocularity has been defined in a variety of ways, and sometimes no definition is given at all. A typical scenario in these studies is that one eye has good vision (usually meaning good acuity and/or visual field sensitivity), whereas the other eye can vary from no vision at all, to acuity worse than 20/200, to significant scotomas in the visual field. With respect to actual driving performance, simulated monocular vision, by occluding one eye, does not appear to impact driving maneuvers on a closed-road course (Wood & Troutbeck, 1992; Wood et al., 1993). Monocular truck drivers also reportedly carried out most maneuvers in a satisfactory fashion (McKnight et al., 1991). With respect to safety, drivers of personal vehicles with monocular field loss did not have an elevated crash rate compared with a control group of drivers with normal visual fields in both eyes (Johnson & Keltner, 1983). However, studies on commercial drivers who have high levels of driving exposure suggest that monocularity, defined as worse than 20/200 in one eye,
elevates crash risk (Laberge-Nadeau et al., 1996; Maag et al., 1997; Rogers, Ratz, & Janke, 1987).
Other Aspects of Vision and Driving
A recent comprehensive review of the color vision and driving literature (Vingrys & Cole, 1988) indicates that color deficiency does not threaten road safety or performance. Color deficiency may pose difficulty in reading traffic control devices in some situations, but the critical cues on the road usually can be obtained through multiple sources of information, allowing drivers to compensate.
Dynamic visual acuity has a stronger unadjusted association to driver safety than does the conventional static acuity test, but the relationship is still weak (Hills & Burg, 1977; Shinar, 1977). Three decades ago, a study showed that performance in a motion perception task was one of the best correlates of self-reported crash involvement among a large battery of vision tests (Shinar, 1977), but motion processing abilities have not received serious examination in the literature in the ensuing years.
Disability glare problems are often discussed as a serious threat to driver safety (Wolbarsht, 1977) but one is hard-pressed to identify actual studies that scientifically confirm this notion. This failure to find an association may be due to methodological difficulties in defining glare and in measuring a multifaceted phenomenon, as well as to a poor understanding of what people mean when they say they have problems with glare.
Direct Measures of Driving Ability
There are no tests of actual driving ability that are widely available, have been standardized, and have proven validity and reliability for use with a wide range of individuals. Licensing agencies have protocols to assess driving performance for their own purposes in order to evaluate applicants for licensure, but these protocols are
highly varied across states and agencies and tailored for the specific needs of each agency and the laws that govern them. On-road driving performance can be evaluated on the open road (i.e., public streets or highways) or on closed courses where other road users and obstacles are nonexistent or minimal. There are several challenges in developing a test of actual driving performance. Driving is a highly complex stream of behaviors, and the practical issue is to decide which ones are the most critical to measure. Second, actual driving involves processing a myriad of events, many of which are rather unexpected; closed courses seriously underestimate the complexity of the driving task because they are sheltered from this reality of the open road. Third, strict standardization of the driving performance evaluation is impossible on the open road because of its uncontrolled nature. Researchers interested in the impact of functional impairments on driving have had some success in developing on-road evaluations that are reliable and valid for cognitively impaired older drivers (Hunt et al., 1993; Odenheimer et al., 1994). However, there is no on-road driving evaluation with demonstrated validity and reliability for drivers of wide-ranging ages who are visually impaired.
An alternative to measuring actual on-road driving performance is to use driving simulators. Performance in simulated driving tasks allows for the evaluation of driving skills in a safe environment, which has heightened relevance when evaluating a driver who has a functional problem, such as visual loss. Simulators enable a controlled testing situation (both stimulus and response) that is easily standardized across testing sessions and drivers. One challenge in the design of simulated driving scenarios is determining the critical aspects of the roadway environment for inclusion in the simulator’s scenes (Ball & Owsley, 1991). Other challenges are sufficient visual fidelity and spatial resolution (e.g., Padmos & Milders, 1992), establishing validity against the gold standard of actual driving performance (e.g., Reinach et al., 1997), and implementing interactive capabilities that sufficiently resemble actual driving. Driving simulators are becoming more popular both for driver training and in research, and these simulators can vary from a highly sophisticated use of computer technology with motion bases (e.g., Iowa Driving Simulator, see Reinach et al., 1997) to part-task simulators that focus on a few critical component skills (Doron simulator). The field of driving simulation is
a rapidly growing field, but at present there is no driving simulator or protocol that has been deemed a standard.
Summary and Recommendations
There are no standard tests of actual driving ability currently available for determining driving fitness in those who are visually impaired. However, over the past few decades, research has identified aspects of vision impairment that elevate crash risk and hamper on-road driving performance. Severe visual field loss in both eyes doubles crash risk, and field constrictions resulting in a less than 40° radius field hamper obstacle avoidance on the road. Severe contrast sensitivity impairment due to age-related cataract (Pelli-Robson scores less than 1.25) elevates crash risk. Slow visual processing speed and divided attention problems also increase crash risk at least twofold; these problems are not detected by visual sensory tests (visual acuity, visual fields, contrast sensitivity). Thus in determining driving fitness, there is a need for a test that screens for these types of visual processing impairments. SSA should support research to develop such tests. Color deficiency does not by itself increase crash risk.
SOCIAL PARTICIPATION
The role of social interaction in a modern, service-oriented economy is very important. In a sample of 2,523 job categories, proportional to the distribution of employed adults in the United States as of 1993 (see Occupational Analysis section for a description of the findings from this database), 47 percent of jobs require a significant degree of advising and instructing, with more than 90 percent indicating the need for routine oral exchange of information. Moreover, in studies of persons with vision loss who describe problems encountered with their impairment, social interactions and activities comprised about 10 percent of the problems encountered (Mangione, Berry, et al., 1998). Social-type problems were the fourth most common type of problem mentioned in this largely older population. Clearly, the
impact of any vision loss on social participation will vary enormously with the individual, the workplace, and the environment in which the individual must function. The manner in which the individual and his or her relatives, friends, coworkers, and the public adapt to the visually based constraints depends on a myriad of complex psychological, social, and workplace-related factors.
There is relatively scant literature on the role of vision in participation in social activities. Social interaction, receiving and acting on cues perceived from others, of course, has been examined in the literature of sociology and psychology, but not in the vision literature. The committee considered the tests of visual function that may be related to social participation (see Chapter 2). In addition, we reviewed the literature on vision-related tests that form specific analogues to social participation. This literature includes two distinct types of studies: those investigating performance-based tests, strongly based in vision, which are presumed to measure some function of social interaction, and those based on self-report of social functioning related to vision.
Performance Tests
The only performance-based tests that have been used to approximate the dimension of social interaction are tests of face recognition. Face recognition is important to personal interactions, but the task is also important for watching TV and movies and being in other audience situations (e.g., congregation, classroom). The task is complex, as multiple cues are used for recognition. Some of these are highly visible features, such as hair color, skin color, and facial hair; others require discerning more subtle detail, such as changes in the shape of the mouth, which may involve shading.
There are multiple versions of face recognition tests; some require the determination of the mood of the face presented, some require identification of well-known persons, and others require identifying a different person embedded in a series of different poses of the same person. The latter version of the test has been shown to have
responses that are sensitive to race, education, and cognitive status apart from visual function (Rubin et al., 1997).
Most studies of visual impairment and the ability to see faces have used images of faces presented at a fixed or limited angular size—often at equivalent viewing distances that simulate a face within arm’s length. The test is also performed typically at ideal light levels with maximum contrast in the photographs displayed, which may not be characteristic of the need to identify faces or read transient facial expressions in a working environment.
Being able to read faces depends in part on viewing distance, which in turn determines the angular size of the face at the viewer’s eye, and there will inevitably be some association with visual acuity. In fact, persons with low vision, primarily represented as central acuity loss or severe visual field loss, report difficulties recognizing familiar faces and discerning facial expression changes (Bullimore et al., 1991). Alexander et al. (1988) studied a large group of subjects, conducting a battery of functional tests that included reading, reading clocks, and distinguishing colors, products, facial expressions. The faces were presented as fixed-size photographs, and it was found that the third of the sample that had the best visual acuity did best at recognizing expressions and the third with the poorest acuity had the worst performance at reading expressions. Bullimore et al. (1991) studied a small group of subjects (n = 13) with macular degeneration. They used photographs of faces and determined the equivalent viewing distances (EVD) required for recognition of identity and expression. These measures of performance were found to be very highly correlated to reading acuity scores and quite strongly correlated to letter chart visual acuity scores. Associations with peak contrast sensitivity and grating acuity were weaker. For some subjects, visual performance was measured under different luminance levels and the within-subject analyses showed parallel changes in the tests of face recognition and the test of reading and grating acuity. They found that the EVDs required for recognition of identity and of expression were very similar except in subjects with very poor visual acuity, for whom recognizing expressions became easier than recognizing identity.
Tests of central acuity and contrast sensitivity have been studied in relation to face recognition and are summarized by Higgins and Bailey (2000). In general, Bailey and colleagues have found a relationship between poor visual acuity, particularly reading acuity, and performance on recognizing facial expressions; modest correlations were also observed with contrast thresholds for sharp edges (Bullimore et al., 1991). However, others found poorer prediction of face recognition from acuity (Rubin & Schuchard, 1989).
There is disagreement on whether low spatial frequency information is all that is required for adequate face recognition, or whether high spatial frequency is also important. There are data to suggest that middle and low spatial frequencies are associated with face recognition (Owsley & Sloane, 1987; Owsley et al., 1981). Work by Peli suggests that the spatial frequency content centered at 8 cycles/face was important for face recognition in those with normal vision, but those with acuity loss preferred images with a center frequency of 16 cycles/ face (Peli, Goldstein, et al., 1991; Peli, Lee, et al., 1994). As Bullimore et al. (1991) point out, the viewing distance for face recognition has an important impact on the influence of acuity versus contrast sensitivity.
In a comprehensive evaluation of vision in a population-based study, the SEE project, the contribution of loss of visual acuity, contrast sensitivity, visual field loss, and stereoacuity has been characterized relative to the decline in the ability to match faces in an older population of 2,520 adults. The test comprised a series of 20 presentations of four faces; in each set of four, three are the same person in different poses, and one is a different person. A full regression model, including adjustments for age, gender, race, education, and cognition, accounted for 37 percent of variance, with the vision variables accounting for 10 percent. Decrements in acuity, contrast sensitivity, visual field, and stereoacuity, independent of each other, were significantly associated with worsening scores in face recognition. While statistically significant, a very small decrement in face recognition was associated with visual field loss and stereoacuity loss, suggesting that these aspects of vision were less important predictors of the score on this test. The parameters suggest that a loss of recognition of one face (unit change) is associated with a three-line
loss in visual acuity and a five-letter loss in contrast sensitivity. There was also an interaction in the data, which suggested that with 20/60 or better visual acuity in the better eye, contrast sensitivity was a major predictor of decreased ability to recognize faces. For visual acuity worse than 20/60, contrast sensitivity decrements contributed little to the decrease in face recognition (West, unpublished).
Self-Report of Social Interaction
Several quality of life scales include the dimensions of role function and social interaction. Unlike the performance-based tests, however, self-report of difficulties includes not only perceived limitations imposed by vision loss, but also such dimensions as expectations of performance and the use of compensatory strategies for visual loss. Thus, there are differences between the actual performance and the self-report of performance for many activities, including reading (Friedman et al., 1999).
Visual acuity loss has been associated with declines in the social function scales of the NEI VFQ-25 (Broman et al., 2001; Mangione, Berry, et al., 1998) and with limitations in communication and recreations and pastimes in the Sickness Impact Profile (SIP) (Scott et al., 1994). Parrish et al., in a study of people with glaucoma, found that binocular visual field impairment was not highly correlated with decrements in social function as measured by the SF-36 or NEI VFQ, once visual acuity impairment was considered (Parrish et al., 1997). Gutierrez et al. found an association with the VF-14 social function item and visual field status of the better eye, but the effect of concomitant acuity was not considered (Gutierrez et al., 1997). Research has studied groups of people with specific eye diseases, such as optic neuritis or age-related maculopathy, and evaluated the correlations of severity of disease with changes in the NEI VFQ; concomitant analyses using tests of vision in these patients were not done (Cole et al., 2000; Mangione et al., 1999).
In preliminary data from the SEE project, participants were asked if they attended social activities such as church, movies, and restaurants
as often as they wished, and if not, whether the decline was due to vision. For attending movies or going to restaurants, contrast sensitivity decrement was the only vision variable significantly associated with a decline in social function (p = .03). Neither visual acuity impairment nor visual field loss was significantly associated, after also adjusting for gender, race, and education.
Summary and Recommendations
The importance of social interaction as a visually intensive task in the workplace environment is generally acknowledged. The use of an instrument for disability determination that would collect data on self-report of decrements in social interaction, or decrements in a performance-based test such as a test of face recognition, as the measure for this skill is not recommended at this time. First, the test of face recognition has an unknown relationship to the visual tasks involved in social interaction. Moreover, there is no standard test of face recognition, nor general agreement on the testing environment that should be used. The correlation with visual acuity and contrast sensitivity for both self-reported measures and face recognition tests suggests that incorporation of these tests as measures of visual function may capture some of the relevant disability in social interaction. Finally, a myriad of other sources of information (e.g., verbal and aural inputs) are also likely to be as critical (if not more so) for social interaction as is visual function. Therefore, tests of social participation should not have high priority at this time, although they merit reconsideration in the future.
TOOL USE AND MANIPULATION
The successful use and manipulation of hand-held tools is a complex task that varies with such components as the complexity of the tool, the reason for the use of the tool, the manual dexterity of the individual, the extent to which hand-eye coordination is needed to use the tool, and the visual demands of the tasks for which the tool is
used. The myriad types of tools, coupled with the variety of tasks for which the tool is being used, do not allow for easy summarization of this topic. Thus, the impact of vision loss on tool use and manipulation will vary enormously with the tool, the tasks, and the individual.
The use of hand-held tools is widespread in the workplace, as suggested by the frequency of tool use (exclusive of controls or keyboard devices) mentioned as important to the job in a sample of 2,523 jobs proportional to the distribution of employed adults in the United States as of 1993 (see below for a description of the findings from this database). Fully 67 percent of jobs reportedly required more than a little use of hand-held machines or equipment, and 37 to 39 percent reported some use of nonprecision tools or instruments and measuring devices. In recent years, the use of tools such as a computer mouse or other hand-held pointing or input device that provides primarily visual feedback has become common for office workers and others whose work involves computer use.3
The visual demands required for jobs with these tools, as reported in the job sample, were varied. It should be noted that the database used to estimate visual demand was based on reports by job incumbents, supervisors, or job analysts of the need for near and far acuity, depth perception, and color vision for each task. Near acuity capability was reported as the most common demand, with a high proportion reporting that near acuity was of moderate to essential importance for tool use. Depth perception and color vision were less likely to be reported as important across all tool use tasks.
Research on the impact of vision loss, measured using standard tests, on performance involving use of tools in industry could not be located. In the past 10 years, the vision research community has borrowed from work done in the gerontology field to identify tasks of everyday living for which vision may have a significant input. Some
of these tasks involve the use of tools, such as sewing, writing a check, etc. There are many sets of these instrumental activities of daily living (IADL) tasks (Lawton & Brody, 1969; Nagi, 1976; Rosow & Breslau, 1966). There are both self-reports of function on these tasks and some standardized analogues for the tasks that can be performed. The literature contains references to the role of vision loss in self-reported difficulties with daily activities requiring tool use, such as garden tools, repair tools, sewing, and the like (see below); fewer data are available on actual performance of the tasks.
Vision and Performance Tests of Tool Use
Immediately following World War II, there was a great deal of physical ability testing being conducted in industry. This was due, in large part, to the ubiquity of machine manufacturing and assembly line subassembly work. As a result, a number of test batteries were developed to assess manual dexterity for near vision tasks. Representative batteries were the MacQuarrie Test for Mechanical Ability, the O’Connor Finger and Tweezer Dexterity Test, the Purdue Pegboard test, the Crawford Small Parts Dexterity Test, and the Minnesota Rate of Manipulation Test. Each of these test batteries is described in some detail by Guion (1965). These tests require the manipulation of some objects with small tools (e.g., tweezers or screwdrivers) or with one’s fingers. These tests were introduced for the purpose of selecting applicants for industrial positions and have not been used as standard batteries for assessing visual impairment. Nevertheless, there are substantial normative data on these tests demonstrating their relationships (i.e., validity) to typical industrial activities. It would seem logical to adapt tests like these for assessing the consequences of various forms of visual impairment on tool use and manipulation in work settings. In conjunction with selected IADL tasks, such a battery might prove extremely useful in determining the extent to which an individual’s visual impairment may influence safe and effective job performance.
Because there is no standard set of performance-based tests, few studies in this area have been done, and tasks were selected that have
a visual component. One study may examine inserting a plug, using a key, and dialing a telephone (West, Munoz, et al., 1997) while others examine telling time and distinguishing products (Alexander et al., 1988), or threading a needle and using a screwdriver (Owsley, McGwin, et al., 2001). Typically, the time required to perform the task and the quality of the performance are graded for each subject. It should be noted again that performance among visually impaired (and nonimpaired) subjects varies greatly depending on such nonvisual factors as the use of compensatory strategies for using the tool, familiarity with the tool (e.g., a sewing needle), strength, dexterity, etc. The tools used for testing are simple and the tasks uncomplicated, which may or may not bear a resemblance to tool use tasks in industry.
Research from one population-based study of older persons indicates that visual acuity deficit and contrast sensitivity loss independently contribute to declines in performance on these tool-oriented tasks, adjusting for other confounders (West et al., in press). In another study, which combined visual acuity loss and/or visual field deficits into a category of “visual impairment,” there was a correlation between visual impairment and performance on an index that included some tool use items (Haymes et al., 2001). It was not possible from this paper to separate out the tool use items specifically. Visual field deficits were less well correlated with performance than was loss of acuity. In a third study of people with retinitis pigmentosa, visual acuity and contrast sensitivity were the only measures of vision (which included visual fields and electroretinogram) associated with level of difficulty in performing such tasks as using a screwdriver, using a vending machine, and pouring water. Only contrast sensitivity was associated with dialing a phone or writing a check. None of the vision measures was associated with threading a needle. None of the vision measures was adjusted for correlations with the others (Szlyk et al., 2001).
In another study of 342 people enrolled in a longitudinal study of mobility, subjects were timed on completion of several tool use tasks, including using a screwdriver, threading a needle, dialing a number, and inserting a key in a lock (Owsley, McGwin, et al., 2001). Acuity, contrast sensitivity, and scores on a test of visual attention and visual processing (useful field of view) were measured for each person. The
tool use tasks were all significantly related to acuity, while contrast sensitivity was independently associated with threading a needle and using a screwdriver. The useful field of view was associated with dialing a phone number. Age, education, comorbidities, and especially cognitive function were important predictors of performance in addition to vision. However, the contribution of vision to performance on these tasks was relatively low.
At present there are insufficient data that link performance on these tasks, or other measures, to actual tool use in an employment setting to be certain that a standard test battery of tool manipulation would be useful in assessing visual disability. The variation in visual demands, depending on the tool and task, also does not allow easy summarization of a single or multiple visual test that would capture deficits in this area.
Self-Report of Tasks Using Tools
Several scales in research instruments include the dimensions of reported difficulty with IADL tasks explicitly, although the tasks themselves may or may not involve the use of tools. Unlike the performance-based tests, however, self-report of difficulties includes not only perceived limitations imposed by vision loss, but also such dimensions as the impact from other comorbid conditions (such as arthritis) and expectations of performance (drill use by a carpenter versus an occasional user, for example). Thus, there are differences between the actual performance and the self-report of performance for many activities, including reading (Friedman et al., 1999).
Several studies have demonstrated that loss of vision, either as reported or measured using standard tests, is associated with self-reported loss of physical function, using instruments that include IADL tasks (Appollonio et al., 1995; Carabellese et al., 1993; Cassard et al., 1995; Dargent-Molina et al., 1996; Havlik, 1986; Jette & Branch, 1985; LaForge et al., 1992; Mangione, Lee, et al., 1998; Mangione et al., 1995; Rubin et al., 2001; Rudberg et al., 1993; Salive et al., 1994). In the vision-related quality of life scales, there is a domain of physical
function in which tool use for daily activities is one component. Visual acuity loss has been associated with declines in the near acuity scales of the NEI VFQ-25, which contains tasks involving tool use, but it is not explicit (Broman et al., 2001).
Rubin et al. have investigated the association of multiple tests of vision with self-report of visual disability (the Activities of Daily Vision instrument). The instrument includes a near vision scale with questions on using such tools as a screwdriver and ruler and threading a needle (Rubin et al., 2001). Loss of visual acuity, contrast sensitivity, stereoacuity, and sensitivity to glare were independently associated with a score of 70 or below (on a scale of 0-100). In addition to vision, age, race, gender, education, cognition, and comorbid conditions also were associated with decrements in function. Peripheral visual field loss (outer 30°), using an 81-point single-intensity screening test, was not related to loss of self-reported function, but central visual field loss (central 30°) was independently related to loss of function. In a study of 62 people with retinitis pigmentosa, Szlyk et al. found a correlation of self-reported difficulties in several tool use tasks with loss of contrast sensitivity, loss of acuity, and loss of visual field (Szlyk et al., 2001). There was no adjustment for the correlation between the various measures of vision.
Summary and Recommendations
While the committee acknowledges that tool use is an important component of daily activities and in the workplace, there are insufficient data to recommend any battery of performance-based tests that include tools that would determine visual disability for this domain. Data suggest that visual acuity loss and contrast sensitivity loss in particular are related to both self-report of difficulty and slower performance using tools.
There are standard, performance-based tests of tool use that appear to be referable to industrial tool use, and we recommend that these should be studied for possible utility in helping to determine disability due to vision loss.
HEALTH-RELATED QUALITY OF LIFE
Health-related quality of life (HRQOL) is a reflection of how a person perceives and reacts to his or her health status as it relates to functioning and well-being. Vision may be an important component of HRQOL. It not only provides sensory input about the surroundings but also influences emotional well-being by enabling performance of activities essential for daily physical and social functioning. Indeed, self-reported difficulties with vision, such as trouble seeing or frequency of blurred vision, are associated with decrements in both physical and social functioning (Lee et al., 1997; Kington et al., 1997). In the social model of disability, visual impairment may thus be associated with undesirable social consequences.
In assessing the relationship between limitations in vision and the ability of individuals to participate in work, we use the general schema of the American Medical Association (AMA) guides for our conceptual framework. At one level, visual impairments are related directly to limitations in an individual’s ability to do a job and to perform important tasks as well as to his or her mental health related to health and vision. At more basic levels, impaired vision functioning can be measured by performance on intermediate surrogates, such as reading speed or tool manipulation. Even more basic are measures that assess only visual and integrative vision function, such as visual field or visual acuity, which have traditionally been used as markers of more complex levels of visual performance. At still more basic levels, electrophysiological tests, such as electroretinograms or visual evoked potentials, assess the physiological function of the visual system.
People perceive and react differently to similar levels of diminished vision, and these perceptions and reactions may, in turn, affect functioning. The committee thus specifically sought to investigate issues surrounding HRQOL. As part of the inquiry into person-centered perspectives on assessment, we investigated the applicability of direct measures of vision-related functional status and health-related quality of life to the determination of vision-related disability, rather than or in addition to using intermediate surrogates.
There are two general strategies for characterizing a person’s state of health or functional ability: objectively measuring it (with varying degrees of input from the individual) and relying on the person to report about it. For example, the traditional tests to assess visual function, even visual fields, are mostly objective physical measures. In contrast, self-ratings of functioning and well-being may be assessed by use of HRQOL questionnaires that are completed by the individual. Information derived from these questionnaires often differs from and may augment information derived from more objective measures.
These questionnaires typically consist of items, scales, and domains. An item is a single question. A scale contains the available categories for expressing a response to the question. A domain identifies a particular focus of attention, such as social functioning, physical functioning, mental health, etc., often represented by several items. Graded responses are required of each item, and a value is assigned to each response. Summation of values and linear transformation results in a subscale score (e.g., 0 to 100) for each domain. In addition to the individual domain scores, many instruments also provide a composite score.
Generic HRQOL Instruments
Instruments that measure HRQOL can be broadly divided into two classes: generic and disease- (e.g., vision-) specific. Generic instruments assess general concepts of health and well-being that are applicable across a range of different types of diseases, medical interventions, and population groups. In the context of visually impaired people, these instruments provide information beyond the immediate impact of a particular visual disorder or symptom; they identify and quantify the impact on wider aspects of daily living, such as self-care, mobility, and dependence. A desirable feature of generic instruments is that they permit comparison of quality of life among different diseases. For example, the impact of having “trouble seeing” has been compared with the impact of diabetes mellitus through the use of a standardized scoring metric (Lee et al., 1997). A weakness of generic instruments is that they often fail to detect subtle changes in specific aspects of
quality of life related to a specific organ or functional system, such as those related to vision and vision impairment (Fryback et al., 1993). Serious medical problems are correlated with lower scores for the generic instruments. Thus, the presence of comorbid conditions must be ascertained and appropriate adjustments made.
Although a number of generic measures of HRQOL have been used for people with visual disorders, the Medical Outcomes Study Form-36 (SF-36) and the Sickness Impact Profile have been most popular. The SF-36 has been used to determine the impact on functional status and well-being of diminished vision (Mangione et al., 1994; Scott et al., 1999), age-related maculopathy (Mangione et al., 1999), glaucoma (Gutierrez et al., 1997; Parrish et al., 1997; Wilson et al., 1998), corneal transplantation (Musch et al., 1997), and treatment for choroidal melanoma (Cruickshanks et al., 1999). The SF-36 was used as the measure of general health-related quality of life in the National Institutes of Health-funded Ocular Hypertension Treatment Study and the Collaborative Longitudinal Evaluation of Kerataconus study (CLEK). The Sickness Impact Profile has been used to study quality of life for people with retinal vascular disease (Scott et al., 1994) and cataracts (Steinberg et al., 1994). Modified for the effects of glaucoma and glaucoma treatment, it was included as one of several questionnaires that were used to assess the quality of life in the Collaborative Initial Glaucoma Treatment Study, another clinical trial sponsored by the National Institutes of Health (Janz et al., 2001).
Studies of the various ophthalmic diseases and their association with scores of these generic health-related quality of life measures have not yielded consistent, unequivocal results. In the few studies in which a statistically significant association was found, the strength of the association was weak.
Vision-Specific HRQOL Instruments
Disease-specific instruments target specific symptoms and problems that are of greatest relevance to a given disease. By assessing health status issues that are more immediately affected, instruments of this
type are typically more sensitive than generic instruments in evaluating clinically relevant aspects of specific diseases and in detecting changes over time.
The early vision-specific questionnaires focused predominantly or exclusively on issues related to diminished vision as a result of cataracts. The most widely used of these have been the Activities of Daily Vision Scale (ADVS) (Mangione et al., 1992) and the VF-14 (Steinberg et al., 1994). The ADVS is a reliable and valid measure of people’s perceptions of functional visual impairment. The VF-14 is similar to the ADVS in content and differs mainly in that there are 14 visual activities of interest rather than 20. (Although these tests may not meet the strictest definition of health-related quality of life instruments because of their exclusive focus on visual functioning, they are popularly used and described as vision-related quality of life instruments.)
Studies using the ADVS and the VF-14 have demonstrated improvement in vision-targeted quality of life after cataract surgery in more than 80 percent of patients (Mangione et al., 1994; Steinberg et al., 1994). It has also been shown that postoperative changes in global satisfaction with vision correlate better with postoperative VF-14 scores than with visual acuity (Steinberg et al., 1994). Data from several studies suggest that vision-targeted quality of life questionnaires predict real-world visual capabilities better than objective measurement of visual acuity (Mangione et al., 1994; Steinberg et al., 1994). For example, a truck driver with problems performing his or her work may be measured as having 20/25 or even 20/20 vision. Whatever visual difficulties are being experienced by the truck driver will not be adequately assessed by a measurement of acuity but will be characterized by a vision-targeted quality of life questionnaire. These results suggest that these questionnaires may in some instances measure visual disability better than routine testing of distance visual acuity with Snellen charts.
The Visual Activities Questionnaire (VAQ) (Sloane et al., 1992) assesses some other dimensions of vision not captured by the ADVS or the VF-14—color discrimination and peripheral vision. This questionnaire was part of a panel of questions that was used to study HRQOL in the
Collaborative Initial Glaucoma Treatment Study. In the VAQ, higher scores (indicating more problems) on the peripheral vision subscale were associated with more visual field loss, but the strength of the correlation was relatively weak (Janz et al., 2001).
The National Eye Institute Visual Function Questionnaire (NEI-VFQ) (Mangione, Lee, et al., 1998) is a recently developed questionnaire that was designed to be relevant to the majority of visually impaired adults, regardless of the cause of the visual disability. Thus it may be more properly called function-specific than disease-specific. Items for the questionnaire were selected after analyses of transcripts from focus groups made up of people with many different ocular diseases. Because it systematically incorporates issues from multiple ocular conditions, the NEI-VFQ may be more appropriate than previous vision-specific instruments for use across a wide range of ophthalmic diseases and impairments. This questionnaire has been used in the assessment of HRQOL in a wide range of diseases, with published results available for people with low vision (Scott et al., 1999), glaucoma (Gutierrez et al., 1997; Parrish et al., 1997), diabetic retinopathy (Klein et al., 2001), and optic neuritis (Cole et al., 2000).
One of the interesting findings thus far reported with the NEI-VFQ is the decline in scores for many of the domains with increased glaucomatous field loss (Gutierrez et al., 1997; Parrish et al., 1997). Furthermore, field loss appears to be independent of visual acuity as a predictor of vision-targeted functioning.
Should Current Measures of HRQOL Be Used in the Disability Determination Process?
The recommendations of previous groups on assessment of visual functioning for disability determination have not addressed the use of health-related quality of life instruments. Although the 1994 Committee on Vision report mentioned the importance of patient perceptions (National Research Council, 1994), the technology and techniques were not sufficiently advanced to allow consideration of such methods. Since that report, substantial new knowledge and
techniques have been developed, allowing us to at least consider the use of HRQOL assessments in disability determination.
The use of HRQOL assessments and instruments has only recently been recognized as an important adjunct to traditional clinical measures. Although these instruments are valid and reliable, it is important to note that they measure self-perceptions. Self-perception of function has many facets that are the product of several factors other than vision. Because of this, the usefulness of these instruments depends in part on the truthfulness of the responses and the lack of influence of other, confounding factors. For example, clinical depression has a very large effect on the reporting of ability to function, apart from any visual problems. Unlike many objective tests in which the responses can be verified, checks on the responses to HRQOL instruments are limited. The validity of a low score could thus be questioned in a situation in which there is incentive to be classified as functionally disabled.
The generic QOL instruments are not likely to be useful for SSA’s purposes. Although it can be argued that data indicating decrement in general health and well-being with some level of diminished vision (acuity and/or visual field) would be helpful, such data currently do not exist. There is some potential in the use of vision-targeted quality of life instruments, particularly the NEI-VFQ. The major advantage of the NEI-VFQ is that its questions are relevant not only to people with cataract but also to those with a wide range of vision-related disorders. However, although studies have documented a decrement in scores (domain-specific and composite) with diminished vision functioning as measured by visual acuity and/or visual field, data relating a specific score to a specific level of functional performance or disability are not available. Thus, at present, a NEI-VFQ score is not translatable to a specific level of visual acuity or visual field impairment.
Summary and Recommendations
There is an emerging consensus that the consideration of HRQOL may provide important information in the assessment of functional
disability. The committee affirms this view. In fact, our research on HRQOL informed and supported our final selection of functional vision domains. However, the value of self-reported data regarding issues of functional impairment due to vision is limited, and the potential for other factors being important and for “gaming” the assessments is great. These considerations led the committee to recommend that health-related quality of life measurements should not be used in the assessment of visual functioning for disability determination.
WORK SKILLS AND VISUAL FUNCTIONING
In approaching the question of how to determine if a worker is “disabled” for visual reasons from performing his or her job, various systems use a variety of surrogate markers. At the most basic level, these surrogates are measures such as visual acuity, visual field, or other psychophysical measures that test the performance of the visual system, although with a required level of participation by the individual. Even more basic levels of assessment, such as electrophysiological testing, that do not require such participation, have not been used to determine disability in adults. As understanding of the often minimal relationship between the traditional surrogates of visual skills and abilities and an individual’s actual abilities to perform important tasks (for example, tool use and reading) has grown, greater attention has been paid to higher levels of surrogates that are more directly and closely related to job skills.
In an effort to better understand the relationships between visual skills and abilities and workplace requirements, the committee pursued several avenues of empirical data collection. First, an analysis of the most common standard vision measures—visual acuity, visual fields, and contrast sensitivity—in relation to actual performance on various tasks such as tool use and reading was conducted using the Salisbury Eye Evaluation study data. Some of these findings were discussed in the sections on social interaction and tool use. Second, a literature review and analysis of the relationship of vision measures—visual acuity and visual field loss—to subjects’ self-reported functioning and
health-related quality of life was conducted and is reported in the preceding section on health-related quality of life instruments.
Third, available labor databases were assessed and analyzed to determine how important facets of vision were related to the performance of job tasks for Department of Labor (DOL) job categories. This third effort was designed to provide a model for determining which visual measures should be weighted more heavily in specific job categories (those in which that visual measure is rated to be more important) and is the subject of this section. It seeks to create a template for more specific and tailored job disability assessments by determining how visual measures should be weighed in determining vision-related disability for each job category and the underlying job tasks that exist within each category. Together, these three efforts will, in the future, allow policy makers to better understand and refine the thresholds for determining vision-related disability for specific job categories.
Job Analysis Using Labor Databases
Job analysis is a process used to examine both the essential functions (i.e., important and frequent tasks) that constitute a job and the knowledge, skills, abilities, and other characteristics that a worker should possess in order to successfully carry out those essential functions. In one form or another, job analysis has been integral to the process of personnel selection since the 1920s. There are two basic types of job analysis, job-oriented and worker-oriented. The job-oriented approach emphasizes the conditions of work, or the results of work. It concentrates more on the accomplishments of work than on the behavior of the worker. The worker-oriented approach “focuses more on the human behaviors that compose the job in question” (Landy, 1989). Currently, more data exists for the use of worker-oriented analyses.
The Position Analysis Questionnaire
Although thousands of worker-oriented job analyses of thousands of job titles have been completed in the past 50 years, most analyses have been completed using idiosyncratic and often poorly defined and inconsistently applied job analysis methods. This makes it very difficult to aggregate the results of those analyses in any meaningful way. Ernest McCormick of Purdue University sought to address these concerns by launching an empirically based project to develop a generic worker-oriented job analysis device that would permit the accumulation of data and the construction of a database for comparing and contrasting job requirements (including such sensory/ perceptual requirements as visual requirements). The result of that effort is the Position Analysis Questionnaire (PAQ), a worker-oriented job analysis system that analyzes jobs based on 187 job elements that describe work activities and work situation variables. The PAQ has been accumulating job analyses since the 1970s.
The development and the validation of the PAQ have been documented by its creators (McCormick et al., 1972). The system itself is well described in technical manuals and technical reports (e.g., McCormick & PAQ Services, Inc., 1977; McCormick et al., 1977). As a result of the process by which the PAQ was developed, validated, and standardized, it is the job analytic device most widely used by employers in the United States. Consequently, the PAQ database includes information on “2000 [now 2,523] jobs that characterize the structure of the U.S. labor force” (McCormick & Jeanneret, 1988). These job titles represent several hundred thousand actual analyses (i.e., multiple analyses of each job title, with the cumulative number of discrete observations varying for each job title from a few to several thousand) done for several thousand different employing organizations. To date, the PAQ has not been used in scholarly analyses, nor has it been used to understand the sensory relationships to job performance in general as opposed to a specified job title.
Department of Labor Occupational Information Network
A second database examined by the committee is the Occupational Information Network (O*NET) system, which provides an online, comprehensive, interactive database system of job descriptions developed by the Employment and Training Administration (ETA) of the U.S. Department of Labor (DOL). The original database used in O*NET 98 (subsequently refined and updated) is based largely on data supplied and refined by occupational analysts from sources such as the Dictionary of Occupational Titles (DOT) (U.S. Employment Service, 1991). Groups of five or six occupational analysts and graduate students in industrial and organizational psychology, working independently, rated each of the occupations using the appropriate survey. O*NET is intended to replace the existing Dictionary of Occupational Titles.
The operational model of O*NET is the Content Model, which includes a questionnaire divided into six sections that represent the major elements for categorizing and classifying job information. The six areas of the Content Model, the characteristics on which each job is rated, are: (1) worker characteristics, (2) worker requirements, (3) occupation requirements, (4) experience requirements, (5) occupation-specific requirements, and (6) occupation characteristics. Under worker characteristics is the category sensory abilities. Under sensory abilities are located seven visual abilities: near vision, far vision, visual color discrimination, night vision, peripheral vision, depth perception, and glare sensitivity (Hubbard et al., 2000). While the structure is promising, O*NET analysis is not yet able to provide meaningful analysis of jobs by worker characteristics. Furthermore, the current O*NET database is not yet a representative sample of the U.S. working population and has not yet been widely used.
Usability of the PAQ Database
The PAQ database includes 2,523 job titles with specific ratings. Each job was assigned to one of the nine DOL aggregate job categories, such as clerical or agriculture and fisheries. On the basis of these nine
larger groupings, the PAQ database was adjusted to be proportional to the distribution of the employed adults in the United States as of 1993. This was accomplished through a weighting of responses to reflect data from the DOL, using the first digit of the DOT code (McCormick et al., 1998b). Therefore, if 3 percent of the jobs were in agriculture and fisheries, then 3 percent of the jobs in the database are in agriculture and fisheries. The PAQ assesses jobs and not the people holding jobs.
In assessing how codes were developed and assigned, our analysis revealed that there are 1,817 unique DOT job codes and the rest are overlapping codes, with data created using a bootstrap technique to take all the persons who responded within that job code and create another entry for that job code, to arrive at the number of 2,523. For each job title, there may have been anywhere from one to several hundred measurement sessions. For each measurement session, there may be one or more observations by different persons (job incumbent, supervisors, consultants, analysts). Because of the proprietary nature of this database, the committee could not determine how many job title assessments are based on only one or very few respondents. It was reported by the developers of the PAQ (Richard Jeanneret, personal communication) that the number is very small. The response in the database for each job title is the average response for all persons who analyzed that job title (thus, the average reponses may include decimals, although the input responses are integers).
There are additional significant limitations in the potential generalizability of the PAQ analysis results. First, the data collected do not arise from a random sample of jobs within each of the 1,817 unique job titles or the nine aggregate categories. Second, and even more importantly, among individuals with specific jobs, the selection of the sample may not be representative of all persons holding that specific job. Furthermore, the responses for each specific job are aggregated over time, starting from the 1970s. As a result, there may be potential biases from changes in jobs that occur over time for which the ratings have not sufficiently reflected the new methods and concerns. This is of particular concern given the averaging of all responses over the past 30 years. For example, 30 years ago, the use of “hand-held control devices” encompassed very different tasks than it
does today, when the use of a computer mouse and other such input devices is included in this category. Similarly, 30 years ago, before computers were common on office desks, most managers would not have reported much use of keyboards.
Other limitations to the PAQ database are inherent in the method of data collection by multiple observers. Individuals who perform the job, their supervisors, and experts in the field may each rate the job. Ratings may therefore vary with individual interpretations of the standards used for the ratings, with raters’ understanding of the visual functions they are rating, and with the raters’ depth of knowledge of the job’s requirements.
Despite these limitations, we did not identify a comparable or better data structure for job analyses in the United States. The validity of the PAQ system has been tested over 30 years by the employer and business communities, which use its results for employment and planning purposes. Reliability ratings suggest that the PAQ has reasonable interrater and test-retest performance (McCormick et al., 1998a). The unique benefits of the PAQ system suggest that it may be used carefully to augment our understanding of vision and vision decrements in the workplace, despite its proprietary nature and the lack of previous use for these purposes.
The PAQ instrument gathers information on the following visual and perceptual characteristics of jobs: near visual acuity, far visual acuity, depth perception, and color discrimination. It does not include assessments of visual field or contrast sensitivity. Furthermore, it measures near vision differently than the other three variables (see below).
Nevertheless, it is possible to enter the database and examine the visual demands in one or all of these categories for specific jobs, job families or categories, or broad occupational groupings. And it is possible to concentrate on the specific visual demands of particular tasks within or across jobs (e.g., the use of hand-held tools, the manipulation of fixed or variable controls, assembly or disassembly tasks). However, because of the data source limitations noted above, the committee chose to use analyses limited only to the nine
TABLE 3-1 Ratings of the Importance to the Job of Aspects of Vision (percentage)
Level of Importance |
Near Acuity |
Far Acuity |
Depth Perception |
Color Perception |
Does not apply |
0.5 |
38.1 |
46.6 |
47.3 |
Very minor |
3.0 |
23.0 |
24.2 |
24.6 |
Low |
17.6 |
19.6 |
14.6 |
13.9 |
Moderate |
69.2 |
13.2 |
9.8 |
9.1 |
High |
8.8 |
5.2 |
4.1 |
3.8 |
Extreme |
0.9 |
1.0 |
0.8 |
1.3 |
Source: Position Analysis Questionnaire database of 2,523 job title ratings. |
aggregate DOL levels and not to use results from variation at the individual job title level.
The PAQ was thus used to determine how important vision is to job category performance in the aggregate and across the nine standard DOL job categories. The first analysis is the distribution of responses to the four main vision items. Respondents were asked to rate the importance to their job of far acuity (defined as differences in seeing characteristics beyond arm’s reach), near acuity (defined as the amount of detail that must be seen within arm’s reach), depth perception (defined as judging the distances between objects), and color perception (defined as differentiating objects or details on the basis of color). As the definitions were worded, only the response for near acuity captures “blind or working in darkness” at the zero end of the scale; the rest of the vision categories score “0” as “does not apply.” Therefore, the distribution for near acuity is much less skewed than for the other vision variables.
As seen in Table 3-1, for the distribution of job titles in the database, far acuity, depth perception, and color perception are of low or minor importance—or do not apply—to 80 percent or more of respondents. However, 69 percent of the respondents rated near acuity as of moderate importance, with 10 percent rating it as of high or extreme importance. While it is clear that the distributions are sensitive to the
structure of the response choices, it is also apparent that the respondents placed considerable importance on near acuity for job performance.
The second analysis is the distribution of responses indicating the importance of four main vision items according to the nine major job categories (Table 3-2). The levels are numerically coded as <3 = low or minor importance; 3 = moderate importance; 4 = high importance;
TABLE 3-2 Rating of the Importance of Vision to Job Performance by Broad Category of Job (percentage)
Vision Importance |
Prof/Tech |
Clerical |
Service |
Ag/Fish |
Near Acuity |
||||
Not Applicable to Low |
11 |
14 |
45 |
51 |
Moderate |
77 |
82 |
46 |
49 |
High |
11 |
4 |
8 |
0 |
Extreme |
1 |
0 |
1 |
0 |
Far Acuity |
||||
Not Applicable to Low |
84 |
93 |
69 |
52 |
Moderate |
11 |
4 |
23 |
24 |
High |
4 |
3 |
5 |
19 |
Extreme |
1 |
0 |
2 |
5 |
Depth Perception |
||||
Not Applicable to Low |
88 |
98 |
82 |
61 |
Moderate |
8 |
2 |
13 |
25 |
High |
4 |
1 |
3 |
11 |
Extreme |
0 |
0 |
1 |
3 |
Color Perception |
||||
Not Applicable to Low |
87 |
94 |
80 |
89 |
Moderate |
9 |
4 |
12 |
11 |
High |
3 |
2 |
7 |
0 |
Extreme |
1 |
1 |
1 |
0 |
(N) |
737 |
707 |
352 |
75 |
Source: Position Analysis Questionnaire database of 2,523 job title ratings. |
5 = extreme importance. Differences by category are apparent. Service jobs and agricultural/fisheries jobs do not rate near acuity as important as do the professional/technical and clerical and sales jobs. However, depth perception is rated more important for jobs in agriculture/fisheries and benchwork, compared with professional and clerical jobs. There is uniformity in the low rating given to the importance of color vision across jobs.
Process |
Machine/Trades |
Bench-work |
Structural |
Misc. |
|
|
|||||
36 |
19 |
12 |
17 |
42 |
|
57 |
67 |
63 |
69 |
46 |
|
5 |
13 |
22 |
12 |
11 |
|
2 |
1 |
3 |
2 |
1 |
|
|
|||||
68 |
78 |
90 |
67 |
56 |
|
26 |
16 |
9 |
20 |
26 |
|
5 |
6 |
1 |
13 |
16 |
|
1 |
0 |
0 |
1 |
2 |
|
|
|||||
77 |
74 |
72 |
69 |
73 |
|
17 |
19 |
14 |
22 |
15 |
|
5 |
6 |
10 |
8 |
9 |
|
1 |
1 |
4 |
1 |
3 |
|
|
|||||
82 |
83 |
72 |
75 |
77 |
|
12 |
11 |
16 |
17 |
16 |
|
6 |
5 |
7 |
6 |
6 |
|
1 |
2 |
5 |
2 |
2 |
|
139 |
191 |
105 |
87 |
110 |
Conclusions
The committee sought to determine if empirical data existed to help understand the relationship between visual skills and abilities and workplace performance and job requirements. Two datasets were identified as being potentially useful. The first, O*NET, was determined to not yet be a representative nor usable data source at this time, although it may become so in the future. The second, the PAQ, was judged to provide information that would otherwise be unavailable.
With the appropriate understanding of the limitation of the PAQ database design, its limitations in execution, and the weighted nature of the data, the PAQ provides information of an otherwise inaccessible nature and was thus used as presented by the committee in a limited fashion. The key limitations of the database reflect in part its proprietary nature and include the following:
-
Representativeness of the U.S. job market (not people) exists only at the aggregate nine-category level, not for any of the 2,523 specific job titles nor across the aggregate of job titles;
-
Each of the 2,523 job titles has an unknown number of measurement sessions and an unknown number of observations per measurement session;
-
The cumulative measurements for each job title are averaged over a 30-year time period and thus may not fully capture changes in workplace requirements as job functions change over time;
-
The PAQ was intended to be used for other purposes than the one the committee is undertaking;
-
The vision variables are limited (no contrast sensitivity and no visual field items) and the near vision variable is measured differently from far vision, color, or depth perception.
The results of our analyses demonstrate that facets of vision remain important to many jobs across the range of job titles or the nine aggregate DOL categories. They also demonstrate, however, that the
importance of vision and of the four measures of visual function being rated—near acuity, far acuity, color, and depth perception—vary across the job categories and specific work or job skills. Because of concerns about the data, we have chosen to limit the analyses to no more detail than the nine categories. We make the following specific points:
-
Analyses of the PAQ database demonstrate that near visual acuity is rated to be of high or extreme importance in 9.7 percent of 2,523 job titles, while far acuity is of similar import in 6.2 percent, depth perception in 4.9 percent, and color perception in 4.1 percent. An additional 69.2 percent of job ratings place near acuity as of moderate importance, while only 13.2 percent so rate far acuity, 9.8 percent depth perception, and 9.1 percent color perception. Thus, the importance of focusing more attention on near acuity assessment for disability determination is suggested.
-
At the same time, the weight to be given to near acuity or to the other vision variables should vary depending on the sector of the economy (i.e., job classification) in question. While 89 percent of professional and technical job ratings indicate that near vision is of at least moderate importance, only 55 percent of service job ratings and 49 percent of agricultural or fisheries job ratings do so. In contrast, 48 percent of agricultural or fisheries job ratings indicate that distance acuity is of at least moderate importance, while only 7 percent of clerical jobs rate it that highly. Depth perception is commonly rated of at least moderate importance only in agricultural and fishing jobs (39 percent) and least important in the clerical sector job ratings (2 percent). Finally, color is most important to the benchwork (28 percent of ratings) and structural (25 percent of ratings) job ratings and least important in the clerical (6 percent).
-
When analyzed by specific job skills relative to each of the nine DOL categories and each of the vision variables, each of these two earlier patterns is seen again. Thus, vision remains relatively important for most job skills, although there is some variation across labor categories and the specific visual task varies by job skill even within the same labor category.
RECOMMENDATIONS FOR TESTS OF VISUAL TASK PERFORMANCE
The committee’s review of the potential for using tests of performance of vision-related tasks determined that for most of the task domains we examined, SSA should not attempt to test performance as a part of its disability determination process, using the tests now available. The one important exception is for reading. The committee also determined that health-related quality of life measures, although useful as indicators of important tasks and activities of daily life that are affected by vision loss, are not appropriate as tests for use in disability determination. The specific recommendations are presented below.
Reading
The committee recommends that a test of reading should be included in the vocational factors steps (Steps 4 and 5) of the disability determination process as soon as a well-normed reading test can be shown to meet test standards to be established by SSA. Reading tests are available that should be able to meet such criteria after modest additional research and development efforts, and we recommend that SSA support such research. We recommend the following criteria for reading tests:
-
Visual characteristics that are consistent with contemporary standards for acuity tests, including a logarithmic progression of print sizes;
-
A range of print sizes containing large enough print to be useful with most visually impaired people and small enough print to reliably measure reading acuity in normally sighted people;
-
Text passages equated in layout across print sizes;
-
Reproducible rules for estimating reading acuity, critical print size, and reading speed;
-
Binocular testing, unless one eye interferes with the other in reading and the problem can be addressed by covering the interfering eye; and
-
Text passages representative in font and letter spacing of commonly encountered real-world texts.
We recommend additional research to establish in more detail the distributions of reading acuity, critical print size, and reading speed in different age groups, and the relationships between these measures and performance of work-related activities and the instrumental activities of daily life.
Mobility
The committee recommends no testing of ambulatory or driving mobility at this time. The evidence suggests that most ambulatory mobility problems will be captured by tests of visual functions: contrast sensitivity, visual fields, and acuity. The committee recommends that mobility problems related to other functions not tested (glare, light adaptation, binocularity, or visuocognitive problems) should be subjectively evaluated for their disabling impact, pending development of acceptable tests of such functions.
For driving mobility, although we recommend no task-based testing at this time, the committee recommends that SSA support development of tests of more complex visual functions that have been shown to affect driving safety, such as divided attention and visual processing speed.
Tool Use and Manipulation
The committee recommends no testing of tool use tasks directly at this time, but we do recommend that SSA support research into adapting commonly used industrial tests that involve tool use for future application to detection of vision-related disability.
Social Participation
The committee recommends no use of any instrument currently available to test the impact of vision loss on social participation. Current research measures, such as tests of face recognition, are not appropriate for such use. Tests of visual acuity and contrast sensitivity should capture some of the relevant disability in social participation. Tests of social participation should not have high priority at this time, although they may merit reconsideration in the future.
Health-Related Quality of Life
The committee’s examination of HRQOL instruments indicates that these instruments can provide valuable information for the assessment of the general relationships of visual impairments to functional disability, but they are not likely to be useful as tests for determining disability in individual claimants. A major weakness is that the instruments rely on self-report of functional status and are thus subject to gaming by claimants motivated to demonstrate loss of function in order to qualify for benefits.