Lovie-Kitchin, Jan E. MSc Optom, PhD, FAAO; Brown, Brian PhD
A number of previous studies have determined the repeatability of high-contrast distance visual acuity (HCDVA), 1–7 but few have examined the repeatability of other common vision measures. Although HCDVA is often considered to be the single most important indicator of the quality of vision, 7 there is recent evidence that other vision measures may provide earlier indicators of vision change. 8, 9 Therefore it would be useful to know the repeatability of other commonly used clinical vision tests, such as low-contrast distance visual acuity (LCDVA), contrast sensitivity (CS), and near visual acuity (NVA), because these may indicate change in visual performance before HCDVA. It is also useful to examine the effect of age on test repeatability. Although it might intuitively be expected that repeatability would worsen with increasing age, Elliott et al. 10 actually showed a slight improvement in repeatability of visual acuity with increased age. Repeatability is known to decrease in low vision patients 11 and in subjects with normal vision under conditions of optical degradation. 12 We ask the question: what is the range of repeatability in subjects with normal visual acuity?
Although there have been examinations of the reliability and intercorrelations of some vision tests, 1, 2, 4–6 to our knowledge there have been no studies in which HCDVA, LCDVA, NVA, and CS have been examined in the same population. Thus, an aim of the current experiment was to extend previous findings to include these measures in the same group of healthy subjects who spanned a wide age range.
We made these measures five times on separate occasions and were thus able to examine normal variability within and between subjects, as well as to examine learning effects on the tests under clinical conditions. Other measures similar to ours (e.g., Reeves et al. 2) have been derived from ‘between subjects’ data. Our data allow significant change to be assessed both for individual patients and for groups. We are also able to assess learning effects by determining if there is a consistent trend of the data toward better scores from visit 1 to visit 5.
Subjects were recruited locally from among the staff of the Queensland University of Technology. Word-of-mouth and e-mail solicitation were used to recruit subjects, so the subjects cannot be considered to be representative of either a campus or a clinic population. The subjects used in this study span a wide age range (from 21 to 68 years) but were not screened for ocular pathology except by self-report. They were selected so that their visual acuity values were within the normal range (Table 1), although their refractive corrections may not have been up to date. Visual acuities with their habitual prescriptions ranged from −0.22 to 0.18 logarithm of the minimum angle of resolution (logMAR). After informed consent had been obtained, subjects attended for five visits, with 2 to 3 days between visits. Ninety-three subjects were initially recruited and 79 completed all five sessions (Table 1).
On the first visit, all measures were made on both eyes; thereafter, measures were made only on the better eye (judged according to HCDVA). The first visit was therefore the longest (no more than 30 min); subsequent visits were no longer than 15 min. Tests were carried out under standardized conditions and in the order HCDVA, LCDVA (using Bailey-Lovie high and low contrast letter charts), CS (using Pelli-Robson charts), and NVA (using Bailey-Lovie near word charts). Distance visual acuities were determined at 3 m; chart luminance was 100 cd/m2; contrast was 86.8% for the high contrast chart and 9.4% for the low contrast chart. The test letters ranged from 6/75 (logMAR 1.1) to 6/2.4 (logMAR −0.4) for both charts at the 3 m test distance. NVA was determined at 40 cm; chart luminance was 136 cd/m2 and contrast was 81.6%. Luminance of the Pelli-Robson chart was 100 cd/m2. All luminances were measured using a Topcon BM-5 photometer.
The Bailey-Lovie distance visual acuity chart 13 for high contrast targets was read until at least three letters on a line were read incorrectly. The final score was calculated by giving credit of −0.02 log units for each letter read. 14 Procedure and scoring for the low contrast chart were identical. In each case, the two available charts were alternated on successive visits.
The Pelli-Robson chart was read at 3 m until two or more errors were made within a triplet of letters; the chart was scored by giving 0.05 contrast sensitivity credit for each letter read correctly. 15 Confusions of C for O were scored as correct. 16 The two sides of the chart were alternated on successive visits.
Bailey-Lovie NVA for high contrast words 17 was determined for each subject. The Bailey-Lovie near vision chart is a high contrast logMAR chart printed in Times font with print sizes from N80 to N2 (10 M to 0.25 M). The chart has two, three, or six unrelated words per line; lines smaller than N24 (6 M) have six words. Word length is four, seven, or 10 letters, and each line with six words has two words of each length. Charts were read at 40 cm, giving equivalent sizes of 1.4 logMAR to −0.2 logMAR. They were read until fewer than three words were correctly read on a line; credit of −0.017 logMAR was given for each word read correctly. Five charts were available, so a different chart was used at each session.
The values from these tests were as expected (Table 2, column 2). The equivalent mean Snellen values for HCDVA, LCDVA, and NVA are 6/5.3, 6/8.6, and 6/6.6, respectively. Mean NVA for these subjects is better than N5 (0.63 M) at 40 cm (despite some less-than-optimal near prescriptions, because many of the patients with presbyopia were ‘between prescriptions’); however, as expected, NVA is not as good as HCDVA (t = −8.73; df = 92; p < 0.001) (Table 2).
Repeated measures allow us to estimate variability in these tests for the subject group as a whole, as well as limits of performance for individual subjects. 3 Bland and Altman 18 have used a plot of the differences of measures as a function of their mean, to estimate variability. If we do this for each of our tests, using the first and fifth measures, we can estimate variability using the SD of these differences. Multiplying the SD of the differences by 1.96 gives the conventional limits of agreement for a particular test, assuming that the distribution of these differences is Gaussian (Table 2, Column 3). This is so for HCDVA, LCDVA, and CS (Kolmogorov-Smirnov test, z ≤ 1.4; p > 0.05), but not for NVA (Kolmogorov-Smirnov test, z = 1.5; p = 0.02).
In addition, because we have five measures of each test for each subject, it is possible to calculate the limit of variability for each subject by multiplying the SD for a given test and subject by 1.96. From these individual variability estimates, we can determine the mean variability for the persons in the group and the range of these values. We have done this for each of the tests, for the group and for persons within the group; these values are also presented in Table 2. Column 4 shows the mean values of this limit, whereas the maximum and minimum values for individual subjects are given in columns 5 and 6, respectively.
For HCDVA, the group data suggest a limit of variability of about one line (0.105 logMAR). To a degree, however, this masks the variability within individual subjects. When their data are examined, it is seen that the mean limit (0.032) is only about one third of the group limit (i.e., only 1.5 to 2 letters). The ‘best’ subject in this group has a limit of less than one letter, whereas the ‘worst’ subject has a variability of almost two lines.
A similar pattern holds true for the other tests; the group limit is greater than the mean individual limit but less than the limit for the most variable subject. The group limits for the HCDVA and LCDVA tests are similar [about one line (Table 2)], and the maximum and minimum individual limits are also similar. The most variable subject has limits of about two lines in each case. NVA limits of performance are close to those of LCDVA with the group limit about one line and 2.5 lines for the subject with most variability; the subject with least variability has a performance limit of only one word (Table 2). A similar pattern again is seen for CS, with much greater limits for the group and the subject with most variability, 3.5 to 4.5 letters, whereas the subject with least variability has a performance limit of less than one letter. The CS group and individual limits are close to those of the other tests, probably because the test is designed to present a logarithmic scale of contrast. Contrast of the retinal image is the factor that ultimately limits discrimination in all visual acuity tests.
There are some practice effects in these data. Practice effects are assessed here as the difference between scores on measure 1 and measure 5 (Fig. 1)—a positive change indicates an improvement. Not surprisingly, on what would be a familiar test to most subjects, the practice effect is minimal (less than one letter on average) for HCDVA, although individual subjects may vary in their scores by ± 1 line. LCDVA shows a greater effect, with improvement of about one letter on average, but these data are more variable than those for HCDVA, with variation of plus or minus two lines. NVA shows a similar improvement with practice, a little under one word, but some subjects show as much as two lines improvement. The CS values show little change on average, but there is substantial variation among individual subjects, with increases and decreases of three to four letters (0.15 to 0.20 log units) in some people.
All of the measured values are influenced by age (Fig. 2), with LCDVA reducing at 1.7 times the rate of HCDVA. NVA shows a more substantial decline, but this may be produced, in part, by the under-corrected presbyopes in this population. Correlations between age and test finding are given in Table 3 only the value for NVA is statistically significant (p < 0.01), but this correlation is suspect because of the inclusion of under-corrected presbyopes. The coefficients of determination are uniformly low, indicating that age is not an important factor in these findings.
Test variability is not greatly influenced by age of the subjects. The standard deviations of the sets of five data points (for 79 subjects) do not vary as a function of age for HCDVA, LCDVA, or CS values (Fig. 3). The fitted regression lines for these measures all have slopes below 0.001 log units per decade, with low and statistically nonsignificant correlation values. The standard deviations of the NVA scores are significantly related to age (r = 0.26, df = 77, p < 0.05), but this is most probably produced by decreased NVA in older subjects, some of whom are not optimally corrected.
As would be expected, scores on these clinical vision tests are related to one another; correlations are highest between the distance visual acuity tests and considerably lower for NVA and CS scores (Table 4). These relationships are also shown in Fig. 4, where the reasonably tight relationship between the two distance visual acuity measures is obvious, but the greater scatter in the other relationships can be appreciated. The relationship between HCDVA and LCDVA compares well with the correlation of HCDVA on measures 1 and 2 for individual subjects (0.81), and the correlation for LCDVA on measures 1 and 2 for individual subjects (0.89).
The relationship between HCDVA and NVA is again contaminated by the lack of up-to-date near prescriptions for some of the older subjects. This would reduce NVA for the subjects with presbyopia; separate analyses of the data of subjects younger and older than 40 years bears this out. For the older subjects, the correlation between HCDVA and NVA is only 0.21 (not significant; n = 43 subjects aged 40 to 68 years) whereas that for younger subjects is 0.65 (p < 0.001; n = 50 subjects aged 21 to 39 years). NVA for all subjects is better than N5 (0.63 M) at 40 cm (0.2 logMAR) but, as indicated above, is significantly worse than HCDVA, which has a mean value of −0.052 logMAR for all measures (t= −8.73; df = 92; p < 0.001). A surprising finding is the relatively low correlation between LCVDA and CS scores (see below). The coefficient of determination for this relationship is 0.24, essentially the same as the value of 0.23 for the relationship between HCVDA and CS scores (Table 4).
Elliott et al. 10 provide an excellent review and discussion of earlier studies of age and visual acuity. They point out that many earlier studies were compromised by use of charts that were truncated (in many cases to 6/6), by use of various refractive conditions (optimal or habitual), or by lack of screening for ocular disease, especially in elderly patients. Only the data of Frisén and Frisén 19 and those of Owsley et al. 20 (for patients younger than 50 years of age) withstood their scrutiny as providing appropriate findings. Frisén and Frisén 19 measured visual acuity in 100 subjects aged from 10 to 75 years. The subjects were optimally corrected and had good general and ocular health. However, they used hand-made charts, which did not conform to standard design principles. The charts were scored letter by letter and frequency-of-seeing curves were fitted by eye; thresholds for 50, 90 and 100% correct reading of the charts were estimated; their data, although useful, cannot be regarded as definitive. However, Elliott et al. 10 do provide definitive data for a large number of optimally corrected subjects with normal ocular health.
The average visual acuity (with habitual prescription) in our subjects was −0.052, with an average loss of 0.01 logMAR per decade of increased age. This compares reasonably well with data from Elliott et al. 10 who measured high contrast Bailey-Lovie visual acuity in 223 subjects ranging from 18 to 80 years in age. They reported mean visual acuity to be −0.13 ± 0.06 in subjects aged 18 to 24 years, declining to −0.02 ± 0.05 in subjects aged 75+ years. Their data suggest a loss of four letters of acuity between the ages of 18 to 24 years and 65 to 69 years (logMAR −0.13 to −0.05). Our data suggest that one letter of acuity is lost for every 20 years of aging, giving a loss of 3.5 letters; these are quite comparable values (despite the fact that our subjects were not optimally corrected). However, the slope of the function fitted to the data of Elliott et al. 10 suggests a loss of 0.029 logMAR per decade (from 30 to 75+ years); in the age range examined here, this slope would produce a loss of almost seven letters. Their data are weighted with older subjects (70 to 74 and 75+ years), whereas we had none in these age groups; this probably accounts for the greater slope of the fitted function. Wood and Bullimore 21 similarly measured high contrast visual acuity using Bailey-Lovie charts for 91 subjects with normal vision in a wider age range (21–82 years); they also found a slope of 0.03 logMAR per decade. The data of Frisén and Frisén 19 for 90% thresholds closely resemble those of Elliott et al. 10 and extend their data at lower ages. Owsley et al. 20 used different charts and different screening criteria for their subjects in younger and older age groups, resulting in a marked difference (0.1–0.2 logMAR) between acuities of these subject groups. Hirvelä et al. 4 measured distance visual acuity in 72 ‘healthy eyes’ of patients older than 70 years and found decimal visual acuity of 1.03 (equivalent to 6/5.8; logMAR −0.016), a value that compares well with the visual acuity values at the high end of our age scale (see Fig. 2). Of particular note in our study, and the recent studies discussed above, is that a large majority of subjects maintain high contrast visual acuity of better than 6/6, well into the older age groups. With improved methods for measuring acuity and for detecting ocular disease, the distinction can be made between early pathological changes and changes with age, which Rowe 22 discriminates as ‘usual aging’ and ‘successful aging’.
Rubin et al. 6 examined a random sample of subjects aged from 65 to 85 years who were optimally corrected and found average visual acuity of −0.01 logMAR for men and 0.0 logMAR for women. Within this age range, they found a loss of 0.018 logMAR per decade, a value almost twice ours. Their data provide evidence for an increased rate of loss of visual acuity in older subjects; however, because a forced choice of responses to control subjects’ response criteria was apparently not used, this cannot give a definitive estimate of visual acuities.
Mean LCDVA of our habitually corrected subjects was 0.16 logMAR. The difference between HCDVA and LCDVA is 0.21 logMAR, just over 2 lines. Such differences have been previously reported; for example, Brown and Lovie-Kitchin 1 reported a difference of 2.5 lines between HCDVA and LCDVA in an uncorrected clinic population aged from 14 to 74 years. Wood and Bullimore 21 also found a difference of 0.2 logMAR for optimally corrected subjects aged 21 to 82 years. Such differences are hypothesized to be caused by scatter in the ocular media or to neural losses in contrast sensitivity at the retina and at more central sites. 23, 24 Abadi and Pantazidou 25 found a loss of about 0.37 logMAR between high- and low-contrast targets on Regan charts in six subjects aged from 54 to 83 years. With a sample as small as theirs, discrepancies may arise; even their youngest subject had a loss of 0.33 logMAR. It is possible that their procedure, of presenting charts in order from lowest to highest contrast (from most difficult to most familiar), biased their result, producing worse acuity for the lower contrast levels. We presented the high-contrast test chart first in all cases.
We measured a rate of decline for LCDVA of 0.017 logMAR per decade of age, which is almost twice that for HCDVA. This value may be low and should be treated with some caution because of the restricted range of both LCDVA and HCDVA values involved in calculating the rate of decline. The relationship between low-contrast visual acuity and age reported by Wood and Bullimore 21 indicate a rate double that which we found (0.036 logMAR per decade); this was clearly influenced by their subjects older than 65 years (see their Fig. 4).
NVA values in our experiment average logMAR 0.044, which is about one line worse than distance visual acuity. Differences of this order have been previously reported for a range of normal and low visual acuities. 14, 26–28
Lovie-Kitchin 28 measured distance and near letter visual acuities with Bailey-Lovie letter charts and near word visual acuity with Bailey-Lovie word charts for a group of 24 subjects with normal vision (optimally corrected), aged 25 to 77 years. Mean distance visual acuity was −0.09 ± 0.14 logMAR, whereas near letter visual acuity was, on average, half a line worse (−0.04 ± 0.11 logMAR). Interestingly, this difference held consistently across each of her low vision groups also. Presumably, variations in accommodation, pupil sizes, and/or depth of focus account for this small difference between distance and near letter visual acuities. Similarly, Lovie-Kitchin 28 found a mean (statistically significant) difference between near word and letter visual acuities of 0.04 ± 0.13 logMAR. This suggests that spatial interaction from letters crowded together in words reduces visual acuity by approximately half a line. Thus it would seem that the different test distances and different stimuli causing variations in accommodation and spatial interactions could explain at least one line (0.1 logMAR) difference between distance letter and near word visual acuities. This agrees very well with the findings in the current study, despite the fact that some subjects were less than optimally corrected.
NVA may be reduced more than distance visual acuity in the presbyopic subjects. This is borne out by the fact that the mean difference between near and distance visual acuity is 0.08 logMAR for those subjects 40 years and younger and 0.13 logMAR for the subjects over 40 years. The rate of decrease of NVA with age is 0.031 logMAR per decade for the whole group of subjects, which is higher than that for either HCDVA or LCDVA. Lovie-Kitchin 28 also found a small but significant decrease in best corrected near word visual acuity of 0.04 logMAR per decade for her normal vision group, which is in close agreement with the current study. Hickson et al. 29 measured Bailey-Lovie near word visual acuity (scored to the nearest whole line) for 240 subjects, aged 60 to 93 years, wearing their habitual near corrections. Eighty-seven percent of subjects had normal NVA, defined as 0.33 logMAR or better [approximately N6 (0.75 M) at 40cm]. As in the current study, they found a significant moderate correlation between age and near VA (r = 0.404), but this would have been confounded by the older age of the subject group and by the increase in the proportion of subjects with visual impairment with increasing age.
CS decreases with increasing age, as might be expected from previous data on the CS function. 20, 30 The average CS of 1.74 that we found for subjects with habitual prescriptions and the decline of 0.016 CS per decade are comparable with data in the literature for studies using the Pelli-Robson test. Elliott et al. 31 reported a CS of 1.92 for young corrected subjects (mean age, 22.5; SD, 4.3 years) and a CS of 1.80 for older corrected subjects (mean age, 70.2; SD, 6.7 years). A decline of 0.027 CS per decade can be estimated from differences between their young and older subjects; however, this is probably an overestimate because of the likely presence of subclinical lens changes in the older subject group. Hirvelä et al. 4 found CS values of 1.69 (SD 0.14) for 72 ‘healthy’ eyes in patients aged over 70 years. As expected, their groups of similar ages with nuclear, cortical or other cataracts, or with macular changes, had worse CS scores.
Wood and Bullimore 21 measured CS for 91 subjects with normal vision, aged 21 to 82 years; CS was significantly correlated with age (r = 0.52 compared with our value of 0.177) and declined at 0.04 CS per decade. The higher fall-off with age that they found is probably attributable the wider age range of subjects that they tested; because of the decrease in CS of older subjects, these subjects would contribute greater losses and increase the slope of the ‘loss function.’
Rubin et al. 6 measured CS (using the Pelli-Robson test) for subjects between 65 and 85 years who were wearing their habitual prescriptions. They found a mean value of about 1.6 (estimated from their Fig. 1); CS was about 1.7 for 65-year-old subjects in this group and the decline of CS with age was 0.12 CS per decade. There was little apparent change in variability with increased age. We found a much lower fall-off in CS with age (0.016 log units/decade); again, this may be because the older subjects in the group used by Rubin et al. 6 had greater prevalence of subclinical lens changes and other media opacities.
Our test-retest reliability data for visual acuity confirm previous findings summarized in Brown and Lovie-Kitchin 3 —the limit for test-retest reliability is about one line for visual acuity tests (if the group is considered), but can be considerably better than that for individual subjects. It may be as low as a few letters for acute observers. There has been a degree of controversy about the limits for change in visual acuity in normal patients, 1, 3, 7, 32 but these data show, as do our previous findings, 3 that one line change is, on average, a useful criterion to adopt in clinical decision making.
This study provides additional data for the repeatability of the Pelli-Robson test, which indicates that variability of three letters should be expected, if working from the group data, but two letters on average if working from the data of individuals. Elliott et al. 31 suggested that six letters (0.3 log units) should be regarded as a significant change, but by scoring the Pelli-Robson chart on a letter-by-letter basis as recommended by Elliott et al., 15 this variability was reduced to about four letters, similar to our findings.
The effects of age in these data are relatively small, although we note that the oldest subject in our sample was only 68 years old. Other studies (e.g., Wood and Bullimore 21 and Rubin et al. 6) have included subjects as old as 85 years, and Haegerstrom-Portnoy et al. 33 investigated habitual vision in elderly subjects aged 58 to 102 years. Our changes per decade for HCDVA, LCDVA, NVA, and CS were 0.01, 0.017, 0.031, and 0.016 log units respectively. Over a ‘normal’ lifespan of 75 years, only 1 to 2 lines (one triplet on the Pelli-Robson chart) can be expected to be lost on any of the charts as a consequence of normal aging. For low-contrast acuity, this loss seems to be present across almost all of the contrast scale, as demonstrated by the (small sample) data of Abadi and Pantazidou, 25 who showed similar losses for letter charts from 4 to 96% contrast. Age-specific losses are also present across 2 log units of luminance as shown by Taub and Sturr, 34 with older subjects having slightly worse performance at low luminance levels. Such losses are also shown with the Smith Kettlewell Low Luminance chart (the SKILL chart) 8, 9; compared with 25-year-olds, 75-year-old subjects lose an additional 0.18 logMAR (nine letters) when tested with the low contrast-low luminance SKILL chart. 9
We saw no increase in variability of test findings with increased age. This concurs with the visual acuity data of Elliott et al. 10 In addition, our data indicate no increase in variability of LCDVA or CS with increased age and these findings also confirm those of Wood and Bullimore. 21
Correlations between Measures
Our findings are in the midrange of correlations (see Table 5), but with coefficients of determination that indicate that < 20% of the variance is accounted for by the correlation. The tests, in the main, are not testing the same underlying functions (except perhaps for the high- and low-contrast visual acuity measures). Because the LCDVA and CS tests both examine aspects of CS, a reasonably high correlation between them might have been expected, but the correlation between them was only 0.49. However, the Pelli-Robson test is said to examine CS at peak spatial frequencies (3–5 c/°), 35 whereas the LCDVA test is probably examining spatial frequencies in the 15–30 c/° range. Care needs to be taken in interpreting these correlations because they could be influenced by other variables such as age or blur.
Hirvelä et al. 4 measured distance visual acuity and CS (using the Cambridge and Pelli-Robson tests) in 72 subjects aged over 70 years ‘with healthy eyes.’ They reported ‘good’ correlations between the CS tests and visual acuity (between 0.47 and 0.79) but did not describe their findings in detail. Rubin et al. 6 reported correlations between visual acuity and Pelli-Robson CS of 0.58, suggesting that 34% of variance between these measures is accounted for by common factors. However, the population used was made up exclusively of subjects aged from 64 to 80 years and included about 5% of subjects with ‘acuity worse than 20/40 in the seeing eye.’
The data presented here indicate that the criterion for judging change on commonly used clinical tests is about one line for each test (one triplet on the Pelli-Robson chart). Although absolute values for test findings reduce as a function of age (by about 2 lines over the age range examined in this study), variability in response does not change, and thus the one line criterion can be applied across all age groups. Testing is more sensitive if the applicable criteria are based on responses of individual subjects rather than those of groups. Adjustments of such criteria, when applied in the absence of specific knowledge of a particular patient, will always be a trade-off between greater sensitivity for detection of change and greater probability of false positives, which will result in either unnecessary additional testing or referral. These findings reinforce the clinical practice of considering all available data and not proceeding based on the findings of a single test.
Our thanks to Cathy Marsh for her assistance with data collection and to Associate Professor Joanne Wood for review of the manuscript.
Supported in part by Australian National Health and Medical Research Grant 951119.
1. Brown B, Lovie-Kitchin JE. High and low contrast acuity and clinical contrast sensitivity tested in a normal population. Optom Vis Sci 1989; 66: 467–73.
2. Reeves BC, Wood JM, Hill AR. Reliability of high- and low-contrast letter charts. Ophthalmic Physiol Opt 1993; 13: 17–26.
3. Brown B, Lovie-Kitchin J. Repeated visual acuity measurement: establishing the patient’s own criterion for change. Optom Vis Sci 1993; 70: 45–53.
4. Hirvelä H, Koskela P, Laatikainen L. Visual acuity and contrast sensitivity in the elderly. Acta Ophthalmol Scand 1995; 73: 111–5.
5. Simpson TL, Regan D. Test-retest variability and correlations between tests of texture processing, motion processing, visual acuity, and contrast sensitivity. Optom Vis Sci 1995; 72: 11–6.
6. Rubin GS, West SK, Munoz B, Bandeen-Roche K, Zeger S, Schein O, Fried LP. A comprehensive assessment of visual impairment in a population of older Americans. The SEE Study. Salisbury Eye Evaluation Project. Invest Ophthalmol Vis Sci 1997; 38: 557–68.
7. Raasch TW, Bailey IL, Bullimore MA. Repeatability of visual acuity measurement. Optom Vis Sci 1998; 75: 342–8.
8. Brabyn J, Schneck M, Haegerstrom-Portnoy G. Vision function of the oldest-old. In: Vision Science and Its Applications, 1996 Technical Digest Series. Washington (DC): Optical Society of America, 1996:14–7.
9. Haegerstrom-Portnoy G, Brabyn J, Schneck ME, Jampolsky A. The SKILL Card. An acuity test of reduced luminance and contrast. Smith- Kettlewell Institute Low Luminance. Invest Ophthalmol Vis Sci 1997; 38: 207–18.
10. Elliott DB, Yang KC, Whitaker D. Visual acuity changes throughout adulthood in normal, healthy eyes: seeing beyond 6/6. Optom Vis Sci 1995; 72: 186–91.
11. Woods RL, Lovie-Kitchin J. The reliability of visual performance measures in low vision. In: Vision Science and Its Applications, 1995 Technical Digest Series. Washington (DC): Optical Society of America, 1995:246–9.
12. Woods RL. Reliability of visual performance measurement under optical degradation. Ophthalmic Physiol Opt 1993; 13: 143–50.
13. Bailey IL, Lovie JE. New design principles for visual acuity letter charts. Am J Optom Physiol Opt 1976; 53: 740–5.
14. Kitchin JE, Bailey IL. Task complexity and visual acuity in senile macular degeneration. Aust J Optom 1981; 64: 235–42.
15. Elliott DB, Bullimore MA, Bailey IL. Improving the reliability of the Pelli-Robson contrast sensitivity test. Clin Vis Sci 1991; 6: 471–5.
16. Elliott DB, Whitaker D, Bonette L. Differences in the legibility of letters at contrast threshold using the Pelli-Robson chart. Ophthalmic Physiol Opt 1990; 10: 323–6.
17. Bailey IL, Lovie JE. The design and use of a new near-vision chart. Am J Optom Physiol Opt 1980; 57: 378–87.
18. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986; 1: 307–10.
19. Frisén L, Frisén M. How good is normal visual acuity?. A study of letter acuity thresholds as a function of age. Albrecht Von Graefes Arch Klin Exp Ophthalmol 1981; 215: 149–57.
20. Owsley C, Sekuler R, Siemsen D. Contrast sensitivity throughout adulthood. Vision Res 1983; 23: 689–99.
21. Wood JM, Bullimore MA. Changes in the lower displacement limit for motion with age. Ophthalmic Physiol Opt 1995; 15: 31–6.
22. Rowe JW, Kahn RL. Human aging: usual and successful. Science 1987; 237: 143–9.
23. Drucker MD, Savino PJ, Sergott RC, Bosley TM, Schatz NJ, Kubilis PS. Low-contrast letter charts to detect subtle optic neuropathies. Am J Ophthalmol 1988; 105: 141–5.
24. Elliott DB. Contrast sensitivity decline with ageing: a neural or optical phenomenon? Ophthalmic Physiol Opt 1987; 7: 415–9.
25. Abadi RV, Pantazidou M. Low contrast letter acuity in age-related maculopathy. Ophthalmic Physiol Opt 1996; 16: 455–9.
26. Sloan LL, Brown DJ. Reading cards for selection of optical aids for the partially sighted. Am J Ophthalmol 1963; 55: 1187–99.
27. Bullimore MA, Bailey IL, Wacker RT. Face recognition in age-related maculopathy. Invest Ophthalmol Vis Sci 1991; 32: 2020–9.
28. Lovie-Kitchin J. Reading performance of adults with low vision. PhD thesis. Queensland University of Technology, 1996.
29. Hickson L, Lind C, Worrall L, Lovie-Kitchin J, Yiu E, Barnett H. Hearing and vision in healthy older Australians: objective and self-report measures. Adv Speech Lang Pathol 1999; 1: 95–105.
30. Crassini B, Brown B, Bowman K. Age-related changes in contrast sensitivity in central and peripheral retina. Perception 1988; 17: 315–32.
31. Elliott DB, Sanderson K, Conkey A. The reliability of the Pelli-Robson contrast sensitivity chart. Ophthalmic Physiol Opt 1990; 10: 21–4.
32. Arditi A, Cagenello R. On the statistical reliability of letter-chart visual acuity measurements. Invest Ophthalmol Vis Sci 1993; 34: 120–9.
33. Haegerstrom-Portnoy G, Schneck ME, Brabyn JA. Seeing into old age: vision function beyond acuity. Optom Vis Sci 1999; 76: 141–58.
34. Taub HA, Sturr JF. The effects of age and ocular health on letter contrast sensitivity and high and medium contrast acuity as a function of luminance. Clin Vis Sci 1991; 6: 181–9.
35. Pelli DG, Robson JG, Wilkins AJ. The design of a new letter chart for measuring contrast sensitivity. Clin Vis Sci 1988; 2: 187–99.